Essential Physics of Medical Imaging

The Essential
Physics of
Medical Imaging
THIRD
EDITION
JERROLD T. BUSHBERG, PhD

Clinical Professor of Radiology and Radiation Oncology
University of California, Davis
Sacramento, California
J. ANTHONY SEIBERT, PhD

Professor of Radiology
EDWIN M. LEIDHOLDT JR, PhD

Clinical Associate Professor of Radiology
JOHN M. BOONE, PhD

Professor of Radiology and Biomedical Engineering
Executive Editor: Charles W. Mitchell

Product Manager: Ryan Shaw
Vendor Manager: Alicia Jackson
Senior Manufacturing Manager: Benjamin Rivera
Senior Marketing Manager: Angela Panetta
Design Coordinator: Stephen Druding
Production Service: SPi Global
Copyright 2012 by LIPPINCOTT WILLIAMS & WILKINS, a WOLTERS KLUWER business
Two Commerce Square
2001 Market Street
Philadelphia, PA 19103 USA
LWW.com
2nd edition 2002 by LIPPINCOTT WILLIAMS & WILKINS
1st edition 1994 by LIPPINCOTT WILLIAMS & WILKINS
All rights reserved. This book is protected by copyright. No part of this book may be reproduced in any form by
any means, including photocopying, or utilized by any information storage and retrieval system without written permission from the copyright owner, except for brief quotations embodied in critical articles and reviews.
Materials appearing in this book prepared by individuals as part of their official duties as U.S. government
employees are not covered by the above-mentioned copyright.
Printed in China
Library of Congress Cataloging-in-Publication Data
Bushberg, Jerrold T.
The essential physics of medical imaging / Jerrold T. Bushberg. 3rd ed.
p. ; cm.
Includes bibliographical references and index.
ISBN 978-0-7817-8057-5
1. Diagnostic imaging. 2. Medical physics. I. Title.
[DNLM: 1. Diagnostic Imagingmethods. WN 200]
RC78.7.D53E87 2011
616.07'54dc22
2011004310
Care has been taken to confirm the accuracy of the information presented and to describe generally accepted
practices. However, the authors, editors, and publisher are not responsible for errors or omissions or for any
consequences from application of the information in this book and make no warranty, expressed or implied,
with respect to the currency, completeness, or accuracy of the contents of the publication. Application of the
information in a particular situation remains the professional responsibility of the practitioner.
The authors, editors, and publisher have exerted every effort to ensure that drug selection and dosage
set forth in this text are in accordance with current recommendations and practice at the time of publication.
However, in view of ongoing research, changes in government regulations, and the constant flow of information
relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for
any change in indications and dosage and for added warnings and precautions. This is particularly important
when the recommended agent is a new or infrequently employed drug.
Some drugs and medical devices presented in the publication have Food and Drug Administration (FDA)
clearance for limited use in restricted research settings. It is the responsibility of the health care provider to
ascertain the FDA status of each drug or device planned for use in their clinical practice.
To purchase additional copies of this book, call our customer service department at (800) 638-3030 or fax
orders to (301) 223-2320. International customers should call (301) 223-2300.
Visit Lippincott Williams & Wilkins on the Internet: at LWW.com. Lippincott Williams & Wilkins customer
service representatives are available from 8:30 am to 6 pm, EST.
10 9 8 7 6 5 4 3 2 1
First and foremost, I offer my most heartfelt love, appreciation and apology to my wife Lori and our children,
Alex and Jennifer, who endured my many absences to focus on completing this text with almost infinite
patience (especially during the last 4 months, when I was typically gone before they woke and got home long
after they had gone to sleep). I look forward to spending much more time with my family and even to starting
to make a dent in the list of chores my wife has been amassing in my absence. I have also had the good fortune
to be supported by my extended family and my Oakshore neighbors who never missed an opportunity to offer
an encouraging word after my response to their question Is the book done yet?
Second, I would like to express my profound gratitude to my coauthors, colleagues, and friends Tony, Ed,
and John for their herculean efforts to bring this 3rd edition into existence. Not only would this text not exist
without them, but the synergy of their combined skills, expertise, and insights was an invaluable resource at
every stage of development of this edition. We all have many more professional obligations now than during the
writing of the previous editions. The willingness and ability of my coauthors to add another substantial commitment of time to their already compressed professional lives were truly remarkable and greatly appreciated.
While all of my staff and colleagues have been very helpful and supportive during this effort (for which I
am very grateful), two individuals deserve special recognition. Linda Krogers willingness to proof read several
chapters for clarity along with the countless other ways she provided her support and assistance during this
effort with her typical intelligent efficiency was invaluable and greatly appreciated. Lorraine Smith has been the
coordinator of our annual radiology resident physics review course for as long as I can remember. This course
would not be possible without her considerable contribution to its success. Lorraine is one of the most helpful,
resourceful, patient, and pleasant individuals I have ever had the pleasure to work with. Her invaluable assistance
with this course, from which this book was developed, is gratefully acknowledged and deeply appreciated.
I would also like to thank our publisher Lippincott Williams and Wilkins, Charley Mitchell, Lisa
McAllister, and in particular Ryan Shaw (our editor) for the opportunity to develop the 3rd edition. Your
patience, support, and firm encouragement to complete this effort are truly appreciated.
I dedicate this edition to my parents. My mother, Annette Lorraine Bushberg (19291981), had a gift for
bringing out the best in me. She cheered my successes, reassured me after my failures, and was an unwavering
source of love and support. My father, Norman Talmadge Bushberg, brightens everyones world with his effortless wit and sense of humor. In addition to his ever present love and encouragement, which have meant more to
me than I can find the words to fully express, he continues to inspire me with his belief in each persons ability
and responsibility to make a unique contribution. To that end, and at the age of 83, he recently published his
first literary contribution, a childrens story entitled Once Upon a Time in Kansas. It is slightly lighter reading
than our text and I highly recommend it. However, if getting your child to fall asleep is the problem, then any
chapter in our book should do the trick.
J.T.B.
Thanks, TSPOON, for your perseverance, patience, and understanding in regard to your often AWOL dad
during these past several yearsits very gratifying to see you prosper in college, and maybe someday you
will be involved in writing a book as well! And to you, Julie Rainwater, for adding more than you know to my
well-being and happiness.
J.A.S.
To my family, especially my parents and my grandmother Mrs. Pearl Ellett Crowgey, and my teachers, especially
my high school mathematics teacher Mrs. Neola Waller, and Drs. James L. Kelly, Roger Rydin, W. Reed Johnson,
and Denny D. Watson of the University of Virginia. To two nuclear medicine physicists, Drs. Mark W. Groch
and L. Stephen Graham, who contributed to earlier editions of this book, but did not live to see this edition.
And to Jacalyn Killeen, who has shown considerable patience during the last year.
E.M.L.
Susan Fris Boone, my wife, makes life on this planet possible and her companionship and support have made my
contribution to this book possible. Emily and Julian, children extraordinaire and both wild travelers of the world,
have grown up using earlier editions of this book as paperweights, lampstands, and coasters. I appreciate the perspective. Marion (Mom) and Jerry (Dad) passed in the last few years, but the support and love they bestowed on
me over their long lives will never be forgotten. Sister Patt demonstrated infinite compassion while nurturing our
parents during their final years and is an angel for all but the wings. Brother Bob is a constant reminder of dedication to patient care, and I hope that someday he and I will both win our long-standing bet. Friends Steve and Susan
have elevated the fun in life. My recent students, Nathan, Clare, Shonket, Orlando, Lin, Sarah, Nicolas, Anita, and
Peymon have helped keep the flag of research flying in the laboratory, and I am especially in debt to Dr. Kai Yang
and Mr. George Burkett who have helped hold it all together during my too frequent travel. There are many more
to thank, but not enough ink. This book was first published in 1994, and over the many years since, I have had
the privilege of sharing the cover credits with my coauthors and good friends Tony, Jerry, and Ed. This has been a
wild ride and it would have been far less interesting if not shared with these tres amigos.
J.M.B.
Preface to the Third Edition

The first edition of this text was written in 1993, and the second edition followed in
2002. This third edition, coming almost 10 years after the second edition, reflects the
considerable changes that have occurred in medical imaging over the past decade.
While the digitization of medical images outside of nuclear medicine began in earnest between the publication of the first and second editions, the transformation of
medical imaging to an all-digital environment is largely complete at the time of this
writing. Recognizing this, we have substantially reduced the treatment of analog modalities in this edition, including only a short discussion on screen-film radiography
and mammography, for example. Because the picture archiving and communication
system (PACS) is now a concrete reality for virtually all radiological image interpretation, and because of the increasing integration between the radiology information
systems (RISs), the PACS, and the electronic medical record (EMR), the informatics
section has been expanded considerably.
There is more to know now than 10 years ago, so we reduced some of the detail
that existed in previous editions that may be considered nonessential today. Detailed
discussions of x-ray tube heating and cooling charts, three-phase x-ray generator
circuits, and CT generations have been shortened or eliminated.
The cumulative radiation dose to the population of the United States from medical imaging has increased about sixfold since 1980, and the use of unacceptably
large radiation doses for imaging patients, including children, has been reported. In
recent years, radiation dose from medical imaging and radiation therapy has become
the focus of much media attention, with a number of radiologists, radiobiologists,
and medical physicists testifying before the FDA and the U.S. Congress regarding the
use of radiation in imaging and radiation therapy. The media attention has given rise
to heightened interest of patients and regulatory agencies in the topics of reporting
and optimizing radiation dose as well as limiting its potentially harmful biological
effects. In this edition, we have added an additional chapter devoted to the topic of
x-ray dose and substantially expanded the chapters on radiation biology and radiation protection. The current International Commission on Radiological Protection
system of estimating the potential detriment (harm) to an irradiated population;
the calculation of effective dose and its appropriate use; as well as the most recent
National Academy of Sciences Biological Effects of Ionizing Radiation (BEIR VII) report
recommended approach of computing radiation risk to a specific individual are discussed in several chapters.
Our publisher has indicated that the second edition was used by increasing numbers of graduate students in medical imaging programs. While the target audience of
this text is still radiologists-in-training, we have added appendices and other sections
with more mathematical rigor than in past editions to increase relevance to scientistsin-training. The goal of providing physicians a text that describes image science and
the radiological modalities in plain English remains, but this third edition contains
an appendix on Fourier transforms and convolution, and Chapter 4 covers basic
image science with some optional mathematics for graduate student readers and for
radiologists with calculus-based undergraduate degrees.
v
vi
Preface to the Third Edition
A number of new technologies that were research projects 10 years ago have
entered clinical use, and this edition discusses the more important of these: tomosynthesis in mammography, cone beam CT, changes in mammography anode composition, the exposure index in radiography, flat panel fluoroscopy, rotational CT
on fluoroscopy systems, iterative reconstruction in CT, and dual modality imaging
systems such as PET/CT and SPECT/CT. Some new technologies offer the possibility
of substantially reducing the radiation dose per imaging procedure.
All of the authors of this book are involved in some way or another with national
or international advisory organizations, and we have added some perspectives from
published documents from the American Association of Physicists in Medicine, the
National Council on Radiation Protection and Measurements, the International Commission on Radiation Units and Measurement, and others.
Lastly, with the third edition we transition to color figures, tables, text headings,
and photographs. Most of the figures are newly designed; some are colorized versions
of figures from previous editions of the text. This edition has been completely rewritten and a small percentage of the text remains as it was in previous editions. We hope
that our efforts on this third edition bring this text to a completely up-to-date status
and that we have captured the most important developments in the field of radiology
so that the text remains current for several years to come.
Foreword
Dr. Bushberg and his coauthors have kept the title The Essential Physics of Medical
Imaging for this third edition. While the first edition in 1994 contained the essentials,
by the time the second edition appeared in 2002, the book had expanded significantly and included not only physics but also a more in depth discussion of radiation
protection, dosimetry, and radiation biology. The second edition became the go to
reference book for medical imaging physics. While not light weekend reading, the
book is probably the only one in the field that you will need on your shelf. Residents
will be happy to know that the third edition contains the topics recommended by the
AAPM and thus likely to appear on future examinations.
Although there are shorter books for board review, those typically are in outline
form and may not be sufficient for the necessary understanding of the topics. This
book is the one most used by residents, medical imaging faculty, and physicists. On
more than one occasion I have heard our university biomedical physicists ask, What
does Bushbergs book say?
The attractive aspects of the book include its completeness, clarity, and ability
to answer questions that I have. This is likely a consequence of the authors having
run a resident review course for almost 30 years, during which they have undoubtedly heard every question and point of confusion that a nonphysicist could possibly
raise. I must say that on the door to my office I keep displayed a quote from the
second edition: Every day there is an alarming increase in the number of things I
know nothing about. Unfortunately, I find this true regarding many things besides
medical physics.
My only suggestion to the authors is that in subsequent editions they delete the
word Essentials from the title, for that word does not do justice to the staggering
amount of work they have done in preparing this editions remarkably clear text or to
the 750+ illustrations that will continue to set the standard for books in this field.
Fred A. Mettler Jr, MD, MPH
Clinical and Emeritus Professor
University of New Mexico School of Medicine
vii
Acknowledgments
During the production of this work, several individuals generously gave their time
and expertise. Without their help, this new edition would not have been possible.
The authors would like to express their gratitude for the invaluable contributions of
the following individuals:
Craig Abbey, PhD
University of California, Santa
Barbara
Ramsey Badawi, PhD
John D. Boice Jr, ScD
Vanderbilt University
Vanderbilt-Ingram Cancer
Center
Michael Buonocore, MD, PhD
Jiang Hsieh, PhD

General Electric Medical Systems
Kiran Jain, MD
Willi Kalender, PhD
Institute of Medical Physics,
Erlangen, Germany
Frederick W. Kremkau, PhD
Wake Forest University School
of Medicine
Fred A. Mettler Jr, MD, MPH

University of New Mexico
School of Medicine
Stuart Mirell, PhD
University of California at Los
Angeles
Norbert Pelc, ScD
Stanford University
Otto G. Raabe, PhD
Werner Roeck, Dipl Eng
University of California, Irvine
Dianna Cody, PhD

MD Anderson Cancer Center
Linda Kroger, MS
Health System
Michael Cronan, RDMS

Ramit Lamba, MD
John Sabol, PhD

General Electric Medical
Systems
Brian Dahlin, MD
Karen Lindfors, MD
D.K. Shelton, MD
University fo California, Davis
Robert Dixon, PhD

Wake Forest University
Mahadevappa Mahesh, PhD

Johns Hopkins University
Jeffrey Siewerdsen, PhD

Johns Hopkins University
Raymond Dougherty, MD
Cynthia McCollough, PhD

Mayo Clinic, Rochester
Michael G. Stabin, PhD

Vanderbilt University
Ken Eldridge, RT(R)(N)
John McGahan, MD
Steve Wilkendorf, RDMS

Sarah McKenney
Sandra Wootton-Gorges, MD
Michael McNitt-Gray, PhD

University of California. Los
Angeles
Kai Yang, PhD

William Erwin, MS
UT MD Anderson Cancer
Center Houston, TX
Kathryn Held, PhD
Massachusetts General Hospital
Harvard Medical School
viii
Contents
Preface to the Third Edition v
Foreword vii
Acknowledgements viii
Section I: Basic Concepts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1 Introduction to Medical Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1 The Modalities 3
1.2 Image Properties 15
Radiation and the Atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.1 Radiation 18
2.2 Structure of the Atom 24
Interaction of Radiation with Matter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.1 Particle Interactions 33
3.2 X-ray and Gamma-Ray Interactions 38
3.3 Attenuation of x-rays and Gamma Rays 44
3.4 Absorption of Energy from X-rays and Gamma Rays 52
3.5 Imparted Energy, Equivalent Dose, and Effective Dose 55
Image Quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.1 Spatial Resolution 60
4.2 Convolution 65
4.3 Physical Mechanisms of Blurring 68
4.4 The Frequency Domain 69
4.5 Contrast Resolution 76
4.6 Noise Texture: The Noise Power Spectrum 86
4.7 Contrast 87
4.8 Contrast-to-Noise Ratio 91
4.9 Signal-to-Noise Ratio 91
4.10 Contrast-Detail Diagrams 92
4.11 Detective Quantum Efficiency 94
4.12 Receiver Operating Characteristic Curves 96
Medical Imaging Informatics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

5.1 Analog and Digital Representation of Data 101
5.2 Digital Radiological Images 109
5.3 Digital Computers 111
5.4 Information Storage Devices 112
5.5 Display of Digital Images 116
5.6 Computer Networks 133
5.7 PACS and Teleradiology 143
5.8 Image Processing 159
5.9 Security, Including Availablility 163
Section II: Diagnostic Radiology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

6 x-ray Production, X-ray Tubes, and x-ray Generators. . . . . . . . . . . . . . . . 171
6.1 Production of x-rays 171
6.2 x-ray Tubes 176
6.3 x-ray Generators 190
6.4 Power Ratings and Heat Loading and Cooling 199
6.5 Factors Affecting x-ray Emission 202
Radiography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
7.1 Geometry of Projection Radiography 207
7.2 Screen-Film Radiography 209
ix
Contents
7.3 Computed Radiography 214
7.4 Charge-Coupled Device and Complementary Metal-Oxide Semiconductor
detectors 217
7.5 Flat Panel Thin-Film-Transistor Array Detectors 220
7.6 Technique Factors in Radiography 223
7.7 Scintillators and Intensifying Screens 224
7.8 Absorption Efficiency and Conversion Efficiency 225
7.9 Other Considerations 226
7.10 Radiographic Detectors, Patient Dose, and Exposure Index 226
7.11 Dual-Energy Radiography 228
7.12 Scattered Radiation in Projection Radiographic Imaging 230
Mammography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
8.1 x-ray Tube and Beam Filtration 240
8.2 x-ray Generator and Phototimer System 250
8.3 Compression, Scattered Radiation, and Magnification 253
8.4 Screen-Film Cassettes and Film Processing 258
8.5 Digital Mammography 263
8.6 Radiation Dosimetry 274
8.7 Regulatory Requirements 276
Fluoroscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
9.1 Functionality 282
9.2 Fluoroscopic Imaging Chain Components 283
9.3 Fluoroscopic Detector Systems 284
9.4 Automatic Exposure Rate Control 292
9.5 Fluoroscopy Modes of Operation 293
9.6 Image Quality in Fluoroscopy 298
9.7 Fluoroscopy Suites 301
9.8 Radiation Dose 304
10 Computed Tomography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312

10.1 Clinical Use 312
10.2 CT System Designs 312
10.3 Modes of CT Acquisition 335
10.4 CT Reconstruction 350
10.5 Image Quality in CT 358
10.6 CT Image Artifacts 367
10.7 CT Generations 370
11 X-ray Dosimetry in Projection Imaging and Computed Tomography. . . . 375

11.1 Attenuation of X-rays in Tissue 375
11.2 Dose-Related Metrics in Radiography and Fluoroscopy 377
11.3 Monte Carlo Dose Computation 382
11.4 Equivalent Dose 383
11.5 Organ Doses from X-ray Procedures 384
11.6 Effective Dose 385
11.7 Absorbed Dose in Radiography and Fluoroscopy 386
11.8 CT Dosimetry and Organ Doses 387
11.9 Computation of Radiation Risk to the Generic Patient 394
11.10 Computation of Patient-Specific Radiation Risk Estimates 396
11.11 Diagnostic Reference Levels 397
11.12 Increasing Radiation Burden from Medical Imaging 399
11.13 Summary: Dose Estimation in Patients 400
12 Magnetic Resonance Basics: Magnetic Fields, Nuclear Magnetic

Characteristics, Tissue Contrast, Image Acquisition. . . . . . . . . . . . . . . . . . 402
12.1 Magnetism, Magnetic Fields, and Magnets 403
12.2 The Magnetic Resonance Signal 412
12.3 Magnetization Properties of Tissues 415
12.4 Basic Acquisition Parameters 420
12.5 Basic Pulse Sequences 421
12.6 MR Signal Localization 438
12.7 K-Space Data Acquisition and Image Reconstruction 444
12.8 Summary 447
Contents
xi
13 Magnetic Resonance Imaging: Advanced Image Acquisition

Methods, Artifacts, Spectroscopy, Quality Control, Siting,
Bioeffects, and Safety. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
13.1 Image Acquisition Time 449
13.2 MR Image Characteristics 460
13.3 Signal from Flow 464
13.3 Perfusion and Diffusion Contrast Imaging 469
13.4 Magnetization Transfer Contrast 473
13.5 MR Artifacts 474
13.6 Magnetic Resonance Spectroscopy 486
13.7 Ancillary Components 488
13.8 Magnet Siting, Quality Control 491
13.9 MR Bioeffects and Safety 495
13.10 Summary 499
14 Ultrasound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
14.1 Characteristics of Sound 501
14.2 Interactions of Ultrasound with Matter 506
14.3 Ultrasound Transducers 513
14.4 Ultrasound Beam Properties 519
14.5 Image Data Acquisition 527
14.6 Two-Dimensional Image Display and Storage 536
14.7 Doppler Ultrasound 542
14.8 Miscellaneous Ultrasound Capabilities 554
14.9 Ultrasound Image Quality and Artifacts 560
14.10 Ultrasound System Performance and Quality Assurance 568
14.11 Acoustic Power and Bioeffects 572
14.12 Summary 575
Section III: Nuclear Medicine. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577

15 Radioactivity and Nuclear Transformation. . . . . . . . . . . . . . . . . . . . . . . . . 579
15.1 Radionuclide Decay Terms and Relationships 579
15.2 Nuclear Transformation 582
16 Radionuclide Production, Radiopharmaceuticals,

and Internal Dosimetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
16.1 Radionuclide Production 594
16.2 Radiopharmaceuticals 608
16.3 Internal Dosimetry 617
16.4 Regulatory Issues 628
17 Radiation Detection andMeasurement. . . . . . . . . . . . . . . . . . . . . . . . . . . 633

17.1 Types of Detectors and Basic Principles 633
17.2 Gas-Filled Detectors 637
17.3 Scintillation Detectors 643
17.4 Semiconductor Detectors 648
17.5 Pulse Height Spectroscopy 651
17.6 Nonimaging Detector Applications 660
17.7 Counting Statistics 667
18 Nuclear ImagingThe Scintillation Camera. . . . . . . . . . . . . . . . . . . . . . . . 674

18.1 Planar Nuclear Imaging: The Anger Scintillation Camera 675
18.2 Computers in Nuclear Imaging 698
19 Nuclear ImagingEmission Tomography. . . . . . . . . . . . . . . . . . . . . . . . . . 705

19.1 Focal Plane Tomography in Nuclear Medicine 705
19.2 Single Photon Emission Computed Tomography 706
19.3 Positron Emission Tomography 720
19.4 Dual Modality ImagingSPECT/CT, PET/CT, and PET/MRI 735
19.5 Clinical Aspects, Comparison of PET and SPECT, and Dose 742
Section IV: Radiation Biology and Protection . . . . . . . . . . . . . . . . . . . . . . . 749

20 Radiation Biology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751
20.1 Overview 751
20.2 Interaction of Radiation with Tissue 752
xii
Contents
20.3 Molecular and Cellular Response to Radiation 757
20.4 Organ System Response to Radiation 772
20.5 Whole Body Response to Radiation: The Acute Radiation Syndrome 784
20.6 Radiation-Induced Carcinogenesis 792
20.7 Hereditary Effects of Radiation Exposure 821
20.8 Radiation Effects In Utero 823
21 Radiation Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 837

21.1 Sources of Exposure to Ionizing Radiation 837
21.2 Personnel Dosimetry 843
21.3 Radiation Detection Equipment in Radiation Safety 850
21.4 Fundamental Principles and Methods of Exposure Control 852
21.5 Structural Shielding of Imaging Facilities 854
21.6 Radiation Protection in Diagnostic and Interventional X-ray Imaging 867
21.7 Radiation Protection in Nuclear Medicine 880
21.8 Regulatory Agencies and Radiation Exposure Limits 892
21.9 Prevention of Errors 897
21.10 Management of Radiation Safety Programs 899
21.11 Imaging of Pregnant and Potentially Pregnant Patients 901
21.12 Medical Emergencies Involving Ionizing Radiation 902
Section V: Appendices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 911

A Fundamental Principles of Physics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 913
B Digital Computers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 929
C Physical Constants, Prefixes, Geometry, Conversion Factors,
and Radiologic Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 938
D Mass Attenuation Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 946
E Effective Doses, Organ Doses, and Fetal Doses from Medical
Imaging Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 955
F Radiopharmaceutical Characteristics and Dosimetry. . . . . . . . . . . . . . . . . 960
G Convolution and Fourier Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 987
H Radiation Dose: Perspectives and Comparisons . . . . . . . . . . . . . . . . . . . . . 998
I
Radionuclide Therapy Home Care Guidelines . . . . . . . . . . . . . . . . . . . . . 1005
Index 1009
Section
Basic Concepts
Chapter
Introduction to Medical Imaging

Medical imaging of the human body requires some form of energy. In the medical
imaging techniques used in radiology, the energy used to produce the image must
be capable of penetrating tissues. Visible light, which has limited ability to penetrate
tissues at depth, is used mostly outside of the radiology department for medical
imaging. Visible light images are used in dermatology (skin photography), gastroenterology and obstetrics (endoscopy), and pathology (light microscopy). Of course,
all disciplines in medicine use direct visual observation, which also utilizes visible
light. In diagnostic radiology, the electromagnetic spectrum outside the visible light
region is used for medical imaging, including x-rays in mammography and computed
tomography (CT); radiofrequency (RF) in magnetic resonance imaging (MRI), and
gamma rays in nuclear medicine. Mechanical energy, in the form of high-frequency
sound waves, is used in ultrasound imaging.
With the exception of nuclear medicine, all medical imaging requires that the
energy used to penetrate the bodys tissues also interacts with those tissues. If
energy were to pass through the body and not experience some type of interaction (e.g., absorption or scattering), then the detected energy would not contain
any useful information regarding the internal anatomy, and thus it would not be
possible to construct an image of the anatomy using that information. In nuclear
medicine imaging, radioactive substances are injected or ingested, and it is the
physiological interactions of the agent that give rise to the information in the
images.
While medical images can have an aesthetic appearance, the diagnostic utility of a medical image relates both to the technical quality of the image and the
conditions of its acquisition. Consequently, the assessment of image quality in
medical imaging involves very little artistic appraisal and a great deal of technical evaluation. In most cases, the image quality that is obtained from medical
imaging devices involves compromisebetter x-ray images can be made when
the radiation dose to the patient is high, better magnetic resonance images can
be made when the image acquisition time is long, and better ultrasound images
result when the ultrasound power levels are large. Of course, patient safety and
comfort must be considered when acquiring medical images; thus, excessive
patient dose in the pursuit of a perfect image is not acceptable. Rather, the power
and energy used to make medical images require a balance between patient safety
and image quality.
1.1 The Modalities

Different types of medical images can be made by varying the types of energies and
the acquisition technology used. The different modes of making images are referred
to as modalities. Each modality has its own applications in medicine.
3
Section I Basic Concepts
Radiography
Radiography was the first medical imaging technology, made possible when the
physicist Wilhelm Roentgen discovered x-rays on November 8, 1895. Roentgen also
made the first radiographic images of human anatomy (Fig. 1-1). Radiography (also
called roentgenography) defined the field of radiology and gave rise to radiologists,
physicians who specialize in the interpretation of medical images. Radiography is
performed with an x-ray source on one side of the patient and a (typically flat) x-ray
detector on the other side. A short-duration (typically less than second) pulse
of x-rays is emitted by the x-ray tube, a large fraction of the x-rays interact in the
patient, and some of the x-rays pass through the patient and reach the detector,
where a radiographic image is formed. The homogeneous distribution of x-rays that
enters the patient is modified by the degree to which the x-rays are removed from the
beam (i.e., attenuated) by scattering and absorption within the tissues. The attenuation properties of tissues such as bone, soft tissue, and air inside the patient are very
different, resulting in a heterogeneous distribution of x-rays that emerges from the
patient. The radiographic image is a picture of this x-ray distribution. The detector
used in radiography can be photographic film (e.g., screen-film radiography) or an
electronic detector system (i.e., digital radiography).
FIGURE 1-1 Wilhelm Conrad Roentgen (18451923) in 1896 (A). Roentgen received the first Nobel Prize
in Physics in 1901 for his discovery of x-rays on November 8, 1895. The beginning of diagnostic radiology is
represented by this famous radiographic image, made by Roentgen on December 22, 1895 of his wifes hand
(B). The bones of her hand as well as two rings on her finger are clearly visible. Within a few months, Roentgen
had determined the basic physical properties of x-rays. Roentgen published his findings in a preliminary report
entitled On a New Kind of Rays on December 28, 1895 in the Proceedings of the Physico-Medical Society
of Wurzburg. An English translation was published in the journal Nature on January 23, 1896. Almost simultaneously, as word of the discovery spread around the world, medical applications of this new kind of ray
rapidly made radiological imaging an essential component of medical care. In keeping with mathematical
conventions, Roentgen assigned the letter x to represent the unknown nature of the ray and thus the term
x-rays was born.
Chapter 1 Introduction to Medical Imaging
Transmission imaging refers to imaging in which the energy source is outside the body
on one side, and the energy passes through the body and is detected on the other side
of the body. Radiography is a transmission imaging modality. Projection imaging refers
to the case when each point on the image corresponds to information along a straightline trajectory through the patient. Radiography is also a projection imaging modality.
Radiographic images are useful for a very wide range of medical indications, including
the diagnosis of broken bones, lung cancer, cardiovascular disorders, etc. (Fig. 1-2).
Fluoroscopy
Fluoroscopy refers to the continuous acquisition of a sequence of x-ray images over
time, essentially a real-time x-ray movie of the patient. It is a transmission projection
imaging modality, and is, in essence, just real-time radiography. Fluoroscopic systems
use x-ray detector systems capable of producing images in rapid temporal sequence.
Fluoroscopy is used for positioning catheters in arteries, visualizing contrast agents
in the GI tract, and for other medical applications such as invasive therapeutic procedures where real-time image feedback is necessary. It is also used to make x-ray movies
of anatomic motion, such as of the heart or the esophagus.
Mammography
Mammography is radiography of the breast, and is thus a transmission projection
type of imaging. To accentuate contrast in the breast, mammography makes use of
FIGURE 1-2 Chest radiography is the most common imaging procedure in diagnostic radiology, often
acquired as orthogonal posterior-anterior (A) and lateral (B) projections to provide information regarding
depth and position of the anatomy. High-energy x-rays are used to reduce the conspicuity of the ribs and other
bones to permit better visualization of air spaces and soft tissue structures in the thorax. The image is a map of
the attenuation of the x-rays: dark areas (high film optical density) correspond to low attenuation, and bright
areas (low film optical density) correspond to high attenuation. C. Lateral cervical spine radiographs are commonly performed to assess suspected neck injury after trauma, and extremity images of the (D) wrist, (E) ankle,
and (F) knee provide low-dose, cost-effective diagnostic information. G. Metal objects, such as this orthopedic
implant designed for fixation of certain types of femoral fractures, are well seen on radiographs.
much lower x-ray energies than general purpose radiography, and consequently the
x-ray and detector systems are designed specifically for breast imaging. Mammography is used to screen asymptomatic women for breast cancer (screening mammography) and is also used to aid in the diagnosis of women with breast symptoms
such as the presence of a lump (diagnostic mammography) (Fig. 1-3A). Digital mammography has eclipsed the use of screen-film mammography in the United States,
and the use of computer-aided detection is widespread in digital mammography.
Some digital mammography systems are now capable of tomosynthesis, whereby the
x-ray tube (and in some cases the detector) moves in an arc from approximately 7 to
40 degrees around the breast. This limited angle tomographic method leads to the
reconstruction of tomosynthesis images (Fig. 1-3B), which are parallel to the plane
of the detector, and can reduce the superimposition of anatomy above and below the
in-focus plane.
Computed Tomography
Computed tomography (CT) became clinically available in the early 1970s, and is
the first medical imaging modality made possible by the computer. CT images are
produced by passing x-rays through the body at a large number of angles, by rotating
the x-ray tube around the body. A detector array, opposite the x-ray source, collects
the transmission projection data. The numerous data points collected in this manner
FIGURE 1-3 Mammography is a specialized x-ray projection imaging technique useful for detecting breast
anomalies such as masses and calcifications. Dedicated mammography equipment uses low x-ray energies,
K-edge filters, compression, screen/film or digital detectors, antiscatter grids and automatic exposure control
to produce breast images of high quality and low x-ray dose. The digital mammogram in (A) shows glandular
and fatty tissues, the skin line of the breast, and a possibly cancerous mass (arrow). In projection mammography, superposition of tissues at different depths can mask the features of malignancy or cause artifacts that
mimic tumors. The digital tomosynthesis image in (B) shows a mid-depth synthesized tomogram. By reducing
overlying and underlying anatomy with the tomosynthesis, the suspected mass in the breast is clearly depicted
with a spiculated appearance, indicative of cancer. X-ray mammography currently is the procedure of choice
for screening and early detection of breast cancer because of high sensitivity, excellent benefit-to-risk ratio,
and low cost.
are synthesized by a computer into tomographic images of the patient. The term
tomography refers to a picture (graph) of a slice (tomo). CT is a transmission technique that results in images of individual slabs of tissue in the patient. The advantage of CT over radiography is its ability to display three-dimensional (3D) slices of
the anatomy of interest, eliminating the superposition of anatomical structures and
thereby presenting an unobstructed view of detailed anatomy to the physician.
CT changed the practice of medicine by substantially reducing the need for
exploratory surgery. Modern CT scanners can acquire 0.50- to 0.62-mm-thick tomographic images along a 50-cm length of the patient (i.e., 800 images) in 5 seconds,
and reveal the presence of cancer, ruptured disks, subdural hematomas, aneurysms,
and many other pathologies (Fig. 1-4). The CT volume data set is essentially isotropic, which has led to the increased use of coronal and sagittal CT images, in addition
to traditional axial images in CT. There are a number of different acquisition modes
available on modern CT scanners, including dual-energy imaging, organ perfusion
imaging, and prospectively gated cardiac CT. While CT is usually used for anatomic
imaging, the use of iodinated contrast injected intravenously allows the functional
assessment of various organs as well.
Because of the speed of acquisition, the high-quality diagnostic images, and the
widespread availability of CT in the United States, CT has replaced a number of imaging procedures that were previously performed radiographically. This trend continues.
However, the wide-scale incorporation of CT into diagnostic medicine has led to more
than 60 million CT scans being performed annually in the United States. This large
number has led to an increase in the radiation burden in the United States, such that
now about half of medical radiation is due to CT. Radiation levels from medical imaging
are now equivalent to background radiation levels in the United States, (NCRP 2009).
FIGURE 1-4 CT reveals superb anatomical detail, as seen in (A) sagittal, (B) coronal, and (C) axial images
from an abdomen-pelvis CT scan. With the injection of iodinated contrast material, CT angiography (CTA)
can be performed, here (D) showing CTA of the head. Analysis of a sequence of temporal images allows
assessment of perfusion; (E) demonstrates a color coded map corresponding to blood volume in this patient
undergoing evaluation for a suspected cerebrovascular accident (stroke). F. Image processing can produce
pseudocolored 3D representations of the anatomy from the CT data.
Magnetic Resonance Imaging

Magnetic resonance imaging (MRI) scanners use magnetic fields that are about
10,000 to 60,000 times stronger than the earths magnetic field. Most MRI utilizes
the nuclear magnetic resonance properties of the protonthat is, the nucleus of the
hydrogen atom, which is very abundant in biological tissues (each cubic millimeter
of tissue contains about 1018 protons). The proton has a magnetic moment and,
when placed in a 1.5 T magnetic field, the proton precesses (wobbles) about its axis
and preferentially absorbs radio wave energy at the resonance frequency of about
64 million cycles per second (megahertzMHz).
In MRI, the patient is placed in the magnetic field, and a pulse of radio waves
is generated by antennas (coils) positioned around the patient. The protons in
the patient absorb the radio waves, and subsequently reemit this radio wave energy
after a period of time that depends upon the spatially dependent magnetic properties of the tissue. The radio waves emitted by the protons in the patient are detected
by the antennas that surround the patient. By slightly changing the strength of the
magnetic field as a function of position in the patient using magnetic field gradients,
the proton resonance frequency varies as a function of position, since frequency is
proportional to magnetic field strength. The MRI system uses the frequency and
phase of the returning radio waves to determine the position of each signal from the
patient. One frequently used mode of operation of MRI systems is referred to as spin
echo imaging.
MRI produces a set of tomographic images depicting slices through the patient, in
which each point in an image depends on the micromagnetic properties of the tissue
corresponding to that point. Because different types of tissue such as fat, white and
gray matter in the brain, cerebral spinal fluid, and cancer all have different local magnetic properties, images made using MRI demonstrate high sensitivity to anatomical
variations and therefore are high in contrast. MRI has demonstrated exceptional utility in neurological imaging (head and spine) and for musculoskeletal applications
such as imaging the knee after athletic injury (Fig. 1-5AD).
MRI is a tomographic imaging modality, and competes with x-ray CT in many
clinical applications. The acquisition of the highest quality images using MRI requires
tens of minutes, whereas a CT scan of the entire head requires seconds. Thus, for
patients where motion cannot be controlled (pediatric patients) or in anatomical
areas where involuntary patient motion occurs (the beating heart and churning intestines), CT is often used instead of MRI. Also, because of the large magnetic field used
in MRI, only specialized electronic monitoring equipment can be used while the
patient is being scanned. Thus, for most trauma, CT is preferred. MRI should not
be performed on patients who have cardiac pacemakers or internal ferromagnetic
objects such as surgical aneurysm clips, metal plate or rod implants, or metal shards
near critical anatomy such as the eye.
Despite some indications for which MRI should not be used, fast image acquisition techniques using special coils have made it possible to produce images in much
shorter periods of time, and this has opened up the potential of using MRI for imaging of the motion-prone thorax and abdomen (Fig. 1-5E). MRI scanners can also
detect the presence of motion, which is useful for monitoring blood flow through
arteries (MR angiographyFigure 1-5F), as well as blood flow in the brain (functional
MR), which leads to the ability to measure brain function correlated to a task
(e.g., finger tapping, response to various stimuli, etc.).
An area of MR data collection that allows for analysis of metabolic products in the
tissue is MR spectroscopy, whereby a single voxel or multiple voxels may be analyzed
FIGURE 1-5 MRI provides excellent and selectable tissue contrast, determined by acquisition pulse
sequences and data acquisition methods. Tomographic images can be acquired and displayed in any plane
including conventional axial, sagittal and coronal planes. (A) Sagittal T1-weighted contrast image of the brain;
(B) axial fluid-attenuated inversion recovery (FLAIR) image showing an area of brain infarct; sagittal image
of the knee, with (C) T1-weighted contrast and (D) T1-weighted contrast with fat saturation (fat signal is
selectively reduced) to visualize structures and signals otherwise overwhelmed by the large fat signal; (E) maximum intensity projection generated from the axial tomographic images of a time-of-flight MR angiogram; (F)
gadolinium contrast-enhanced abdominal image, acquired with a fast imaging employing steady-state acquisition sequence, which allows very short acquisition times to provide high signal-to-noise ratio of fluid-filled
structures and reduce the effects of patient motion.
using specialized MRI sequences to evaluate the biochemical composition of tissues

in a precisely defined volume. The spectroscopic signal can act as a signature for
tumors and other maladies.
Ultrasound Imaging
When a book is dropped on a table, the impact causes pressure waves (called sound)
to propagate through the air such that they can be heard at a distance. Mechanical
energy in the form of high-frequency (ultra) sound can be used to generate images
of the anatomy of a patient. A short-duration pulse of sound is generated by an
10
ultrasound transducer that is in direct physical contact with the tissues being imaged.
The sound waves travel into the tissue, and are reflected by internal structures in the
body, creating echoes. The reflected sound waves then reach the transducer, which
records the returning sound. This mode of operation of an ultrasound device is
called pulse echo imaging. The sound beam is swept over a slice of the patient line by
line using a linear array multielement transducer to produce a rectangular scanned
area, or through incremental angles with a phased array multielement transducer to
produce a sector scanned area. The echo amplitudes from each line of ultrasound
are recorded and used to compute a brightness mode image with grayscale-encoded
acoustic signals representing a tomographic slice of the tissues of interest.
Ultrasound is reflected strongly by interfaces, such as the surfaces and internal
structures of abdominal organs. Because ultrasound is thought to be less harmful
than ionizing radiation to a growing fetus, ultrasound imaging is preferred in obstetrical patients (Fig. 1-6A,B). An interface between tissue and air is highly echoic, and
thus, very little sound can penetrate from tissue into an air-filled cavity. Therefore,
ultrasound imaging has less utility in the thorax where the air in the lungs presents a
FIGURE 1-6 The ultrasound image is a map of the echoes from tissue boundaries of high-frequency sound
wave pulses. A. A phased-array transducer operating at 3.5 MHz produced the normal obstetrical ultrasound
image (sagittal profile) of Jennifer Lauren Bushberg at 5 months before her first birthday. Variations in
the image brightness are due to acoustic characteristics of the tissues; for example, the fluid in the placenta is
echo free, whereas most fetal tissues are echogenic and produce larger returned signals. Acoustic shadowing
is caused by highly attenuating or scattering tissues, such as bone or air, producing the corresponding low
intensity streaks distal to the transducer. B. Distance measurements (e.g., fetal head diameter assessment for
age estimation) are part of the diagnostic evaluation of a cross-sectional brain ultrasound image of a fetus. C.
From a stack of tomographic images acquired with known geometry and image locations, 3D image rendering
of the acoustic image data can show anatomic findings, such as a cleft palate of the fetus. D. Vascular assessment using Doppler color-flow imaging can be performed by many ultrasound systems. A color-flow image of
the internal carotid artery superimposed on the grayscale image demonstrates an aneurysm in the left internal
carotid artery of this patient.
11
barrier that the sound beam cannot penetrate. Similarly, an interface between tissue
and bone is also highly echoic, thus making brain imaging, for example, impractical in most cases. Because each ultrasound image represents a tomographic slice,
multiple images spaced a known distance apart represent a volume of tissue, and
with specialized algorithms, anatomy can be reconstructed with volume rendering
methods as shown in Figure 1-6C.
Doppler Ultrasound
Doppler ultrasound makes use of a phenomenon familiar to train enthusiasts.
For the observer standing beside railroad tracks as a rapidly moving train goes
by blowing its whistle, the pitch of the whistle is higher as the train approaches
and becomes lower as the train passes by the observer and speeds off into the
distance. The change in the pitch of the whistle, which is an apparent change in
the frequency of the sound, is a result of the Doppler effect. The same phenomenon occurs at ultrasound frequencies, and the change in frequency (the Doppler
shift) is used to measure the motion of blood. Both the speed and direction of
blood flow can be measured, and within a subarea of the grayscale image, a color
flow display typically shows blood flow in one direction as red, and in the other
direction as blue. In Figure 1-6D, a color-flow map reveals arterial flow of the left
internal carotid artery superimposed upon the grayscale image; the small, multicolored nub on the vessel demonstrates complex flow patterns of an ulcerated
aneurysm.
Nuclear Medicine Imaging

Nuclear medicine is the branch of radiology in which a chemical or other substance
containing a radioactive isotope is given to the patient orally, by injection or by
inhalation. Once the material has distributed itself according to the physiological
status of the patient, a radiation detector is used to make projection images from
the x- and/or gamma rays emitted during radioactive decay of the agent. Nuclear
medicine produces emission images (as opposed to transmission images), because
the radioisotopes emit their energy from inside the patient.
Nuclear medicine imaging is a form of functional imaging. Rather than yielding
information about just the anatomy of the patient, nuclear medicine images provide information regarding the physiological conditions in the patient. For example,
thallium tends to concentrate in normal heart muscle, but in areas that are infarcted
or are ischemic, thallium does not concentrate as well. These areas appear as cold
spots on a nuclear medicine image, and are indicative of the functional status of
the heart. Thyroid tissue has a great affinity for iodine, and by administering radioactive iodine (or its analogues), the thyroid can be imaged. If thyroid cancer has
metastasized in the patient, then hot spots indicating their location may be present on the nuclear medicine images. Thus functional imaging is the forte of nuclear
medicine.
Nuclear Medicine Planar Imaging

Nuclear medicine planar images are projection images, since each point on the
image is representative of the radioisotope activity along a line projected through the
patient. Planar nuclear images are essentially 2D maps of the 3D radioisotope distribution, and are helpful in the evaluation of a large number of disorders (Fig. 1-7).
12
FIGURE 1-7 Anterior and posterior whole-body bone

scan images of a 64-year-old male with prostate cancer. This patient was injected with 925 MBq (25 mCi) of
99m
Tc methylenediphosphonate (MDP) and was imaged
3 hours later with a dual-head scintillation camera. The
images demonstrate multiple metastatic lesions. Lesions
are readily seen in ribs, sternum, spine, pelvis, femurs
and left tibia. Planar imaging is still the standard for
many nuclear medicine examinations (e.g., whole-body
bone scans and hepatobiliary thyroid, renal and pulmonary studies). (Image courtesy of DK Shelton.)
Single Photon Emission Computed Tomography

Single photon emission computed tomography (SPECT) is the tomographic counterpart of nuclear medicine planar imaging, just like CT is the tomographic counterpart
of radiography. In SPECT, a nuclear camera records x- or gamma-ray emissions from
the patient from a series of different angles around the patient. These projection data
are used to reconstruct a series of tomographic emission images. SPECT images provide diagnostic functional information similar to nuclear planar examinations; however, their tomographic nature allows physicians to better understand the precise
distribution of the radioactive agent, and to make a better assessment of the function
of specific organs or tissues within the body (Fig. 1-8). The same radioactive isotopes
are used in both planar nuclear imaging and SPECT.
Positron Emission Tomography

Positrons are positively charged electrons, and are emitted by some radioactive
isotopes such as fluorine-18 and oxygen-15. These radioisotopes are incorporated
into metabolically relevant compounds, such as 18F-fluorodeoxyglucose (18FDG),
which localize in the body after administration. The decay of the isotope produces
a positron, which rapidly undergoes a very unique interaction: the positron (e)
combines with an electron (e) from the surrounding tissue, and the mass of both
the e and the e is converted by annihilation into pure energy, following Einsteins
famous equation E mc2. The energy that is emitted is called annihilation radiation.
Annihilation radiation production is similar to gamma ray emission, except that
two photons are produced, and they are emitted simultaneously in almost exactly
opposite directions, that is, 180 degrees from each other. A positron emission
tomography (PET) scanner utilizes rings of detectors that surround the patient, and
has special circuitry that is capable of identifying the photon pairs produced during
13
FIGURE 1-8 Two-day stress-rest myocardial perfusion imaging with SPECT/CT was performed on an
9-year-old, obese male with a history of prior CABG, bradycardia, and syncope. This patient had pharmaco8
logical stress with regadenoson and was injected with 1.11 GBq (30 mCi) of 99mTc-tetrofosmin at peak stress.
Stress imaging followed 30 minutes later, on a variable-angle two-headed SPECT camera. Image data were
acquired over 180 degrees at 20 seconds per stop. The rest imaging was done 24 hours later with a 1.11 GBq
(30 mCi) injection of 99mTc-tetrofosmin. Stress and rest perfusion tomographic images are shown on the left
side in the short axis, horizontal long axis, and vertical long axis views. Bullseye and 3D tomographic images
are shown in the right panel. Stress and rest images on the bottom (IRNC) demonstrate count reduction in the
inferior wall due to diaphragmatic attenuation. The same images corrected for attenuation by CT (IRAC) on the
top better demonstrate the inferior wall perfusion reduction on stress, which is normal on rest. This is referred
to as a reversible perfusion defect which is due to coronary disease or ischemia in the distribution of the
posterior descending artery. SPECT/CT is becoming the standard for a number of nuclear medicine examinations, including myocardial perfusion imaging. (Image courtesy of DK Shelton.)
a nnihilation. When a photon pair is detected by two detectors on the scanner, it

is assumed that the annihilation event took place somewhere along a straight line
between those two detectors. This information is used to mathematically compute
the 3D distribution of the PET agent, resulting in a set of tomographic emission
images.
Although more expensive than SPECT, PET has clinical advantages in certain diagnostic areas. The PET detector system is more sensitive to the presence of radioisotopes
than SPECT cameras, and thus can detect very subtle pathologies. Furthermore, many
of the elements that emit positrons (carbon, oxygen, fluorine) are quite physiologically
relevant (fluorine is a good substitute for a hydroxyl group), and can be incorporated
into a large number of biochemicals. The most important of these is 18FDG, which is
concentrated in tissues of high glucose metabolism such as primary tumors and their
metastases. PET scans of cancer patients have the ability in many cases to assess the
extent of disease, which may be underestimated by CT alone, and to serve as a baseline against which the effectiveness of chemotherapy can be evaluated. PET studies are
often combined with CT images acquired immediately before or after the PET scan.
14
PET/CT combined imaging has applications in oncology, cardiology, neurology,

and infection and has become a routine diagnostic tool for cancer staging. Its role
in the early assessment of the effectiveness of cancer treatment reduces the time,
expense, and morbidity from failed therapy (Fig. 1-9).
Combined Imaging Modalities

Each of the imaging modalities has strengths (e.g., very high spatial resolution in
radiography) and limitations (e.g., anatomical superposition in radiography). In
particular, nuclear medicine imaging, whether with a scintillation camera or PET,
often shows abnormalities with high contrast, but with insufficient anatomic
detail to permit identification of the organ or tissue with the lesion. Furthermore,
in nuclear medicine, attenuation by the patient of emitted radiation degrades
the information in the images. Combining a nuclear medicine imaging system
(SPECT or PET) with another imaging system providing good definition of anatomy (CT or MRI) permits the creation of fused images, enabling anatomic localization of abnormalities, and correction of the emission images for attenuation
(Fig. 1-10).
FIGURE 1-9 Two whole-body PET/CT studies of a 68-year-old male undergoing treatment for small cell lung
cancer. Maximal intensity projection images are shown before (A) and after (B) chemotherapy. For each study,
the patient was injected intravenously with 740 MBq (20 mCi) of 18FDG. The CT acquisition was immediately
followed by the PET study, which was acquired for 30 minutes, beginning 60 minutes after injection of the
FDG. The bottom panel shows the colorized PET/CT fusion axial images before (C) and after (D) chemotherapy.
The primary tumor in the right hilum (C) is very FDG avid (i.e., hypermetabolic). The corresponding axial slice
(D) was acquired 3 months later, showing dramatic metabolic response to the chemotherapy. The metastatic
foci in the right scapula and left posterior rib have also resolved. The unique abilities of the PET/CT scan in this
case were to assess the extent of disease, which was underestimated by CT alone, and to assess the effectiveness of chemotherapy. (Images courtesy of DK Shelton.)
15
FIGURE 1-10 A planar and SPECT/CT bone scan done 3 hours after injection of 925 MBq (25 mCi) Tc-MDP.
A. Anterior (left) and posterior (right) spot views of the spine in this 54-year-old female with back pain. The
posterior view shows a faintly seen focus over a lower, right facet of the lumbar spine. B. Coronal views of the
subsequent SPECT bone scan (left) better demonstrate the focus on the right lumbar spine at L4. The colorized image of the SPECT bone scan with CT fusion is shown on right. C. The axial views of the SPECT bone
scan (left) and the colorized SPECT/CT fusion image (right) best localizes the abnormality in the right L4 facet,
consistent with active facet arthropathy. (Images courtesy of DK Shelton.)
1.2 Image Properties
Contrast
Contrast in an image manifests as differences in the grayscale values in the image.
A uniformly gray image has no contrast, whereas an image with sharp transitions between
dark gray and light gray demonstrates high contrast. The various imaging modalities
introduced above generate contrast using a number of different forms of energy, which
interact within the patients tissues based upon different physical properties.
The contrast in x-ray transmission imaging (radiography, fluoroscopy, mammography, and CT) is produced by differences in tissue composition, which determine
the local x-ray absorption coefficient, which in turn is dependent upon the density
(g/cm3) and the effective atomic number. The energies of the x-ray photons in the
beam (adjusted by the operator) also affect contrast in x-ray images. Because bone
has a markedly different effective atomic number (Zeff < 13) than soft tissue (Zeff < 7),
due to its high concentration of calcium (Z 20) and phosphorus (Z 15), bones
16
produce high contrast on x-ray image modalities. The chest radiograph, which demonstrates the lung parenchyma with high tissue and airway contrast, is the most
common radiographic procedure performed in the world (Fig. 1-2).
CTs contrast is enhanced over other x-ray imaging modalities due to its tomographic nature. The absence of out-of-slice structures in the CT image greatly
improves its image contrast.
Nuclear medicine images (planar images, SPECT, and PET) are maps of the spatial distribution of radioisotopes in the patient. Thus, contrast in nuclear images
depends upon the tissues ability to concentrate the radioactive material. The uptake
of a radiopharmaceutical administered to the patient is dependent upon the pharmacological interaction of the agent with the body. PET and SPECT have much better
contrast than planar nuclear imaging because, like CT, the images are not obscured
by out-of-slice structures.
Contrast in MR imaging is related primarily to the proton density and to relaxation phenomena (i.e., how fast a group of protons gives up its absorbed energy).
Proton density is influenced by the mass density (g/cm3), so MRI can produce images
that look somewhat like CT images. Proton density differs among tissue types, and
in particular adipose tissues have a higher proportion of protons than other tissues,
due to the high concentration of hydrogen in fat (CH3(CH2)nCOOH). Two different relaxation mechanisms (spin/lattice and spin/spin) are present in tissue, and
the dominance of one over the other can be manipulated by the timing of the RF
pulse sequence and magnetic field variations in the MRI system. Through the clever
application of different pulse sequences, blood flow can be detected using MRI techniques, giving rise to the field of MR angiography. Contrast mechanisms in MRI are
complex, and thus provide for the flexibility and utility of MR as a diagnostic tool.
Contrast in ultrasound imaging is largely determined by the acoustic properties
of the tissues being imaged. The difference between the acoustic impedances (tissue
density speed of sound in tissue) of two adjacent tissues or other substances affects
the amplitude of the returning ultrasound signal. Hence, contrast is quite apparent at tissue interfaces where the differences in acoustic impedance are large. Thus,
ultrasound images display unique information about patient anatomy not provided
by other imaging modalities. Doppler ultrasound imaging shows the amplitude and
direction of blood flow by analyzing the frequency shift in the reflected signal, and
thus, motion is the source of contrast.
Spatial Resolution
Just as each modality has different mechanisms for providing contrast, each modality
also has different abilities to resolve fine detail in the patient. Spatial resolution refers
to the ability to see small detail, and an imaging system has higher spatial resolution
if it can demonstrate the presence of smaller objects in the image. The limiting spatial
resolution is the size of the smallest object that an imaging system can resolve.
Table 1-1 lists the limiting spatial resolution of each of the imaging modalities
used in medical imaging. The wavelength of the energy used to probe the object
is a fundamental limitation of the spatial resolution of an imaging modality. For
example, optical microscopes cannot resolve objects smaller than the wavelengths
of visible light, about 400 to 700 nm. The wavelength of x-rays depends on the
x-ray energy, but even the longest x-ray wavelengths are tinyabout 1 nm. This is
far from the actual resolution in x-ray imaging, but it does represent the theoretical
limit on the spatial resolution using x-rays. In ultrasound imaging, the wavelength
of sound is the fundamental limit of spatial resolution. At 3.5 MHz, the wavelength
of sound in soft tissue is about 500 mm. At 10 MHz, the wavelength is 150 mm.
17
TABLE 1-1THE LIMITING SPATIAL RESOLUTIONS OF VARIOUS MEDICAL
IMAGING MODALITIES. THE RESOLUTION LEVELS ACHIEVED IN

TYPICAL CLINICAL USAGE OF THE MODALITY ARE LISTED
MODALITY
SPATIAL RESOLUTION (mm)
COMMENTS
Screen film radiography
0.08
Limited by focal spot size and

detector resolution
Digital radiography
0.17
Limited by size of detector elements

and focal spot size
Fluoroscopy
0.125
Limited by detector resolution and

focal spot size
Screen film mammography
0.03
Highest resolution modality in

radiology, limited by same factors
as in screen film radiography
Digital mammography
0.050.10
Limited by same factors as digital

radiography
Computed tomography
0.3
About mm pixels
Nuclear medicine planar

imaging
2.5 (detector face),

5 (10 cm from detector)
Spatial resolution degrades

substantially with distance from
detector
Single photon emission

computed tomography
Spatial resolution worst towards

the center of cross-sectional image
slice
Positron emission tomography 5
Better spatial resolution than the

other nuclear imaging modalities
Magnetic resonance imaging
1.0
Resolution can improve at higher

magnetic fields
Ultrasound imaging (5 MHz)
0.3
Limited by wavelength of sound
MRI poses a paradox to the wavelength-imposed resolution rulethe wavelength of

the radiofrequency waves used (at 1.5 T, 64 MHz) is 470 cm, but the spatial resolution
of MRI is better than a millimeter. This is because the spatial distribution of the paths of
RF energy is not used to form the actual image (contrary to ultrasound, light microscopy,
and x-ray images). The radiofrequency energy is collected by a large antenna, and it carries the spatial information of a group of protons encoded in its frequency spectrum.
Medical imaging makes use of a variety of physical parameters as the source of
image information. The mechanisms for generating contrast and the spatial resolution
properties differ amongst the modalities, thus providing a wide range of diagnostic
tools for referring physicians. The optimal detection or assessment of a specific clinical
condition depends upon its anatomical location and tissue characteristics. The selection of the best modality for a particular clinical situation requires an understanding
of the physical principles of each of the imaging modalities. The following chapters of
this book are aimed at giving the medical practitioner just that knowledge.
SELECTED REFERENCE
NCRP 2009: National Council on Radiation Protection and Measurements. Ionizing radiation exposure
of the population of the United States (NCRP Report No 160). Bethesda, Md: National Council on
Radiation Protection and Measurements; 2009.
Chapter
Radiation and the Atom

2.1
Radiation
Radiation is energy that travels through space or matter. Two catogories of radiation
of importance in medical imaging are electromagnetic (EM) and particulate.
Electromagnetic Radiation
Radio waves, visible light, x-rays, and gamma rays are different types of EM radiation.
EM radiation has no mass, is unaffected by either electric or magnetic fields, and has a
constant speed in a given medium. Although EM radiation propagates through matter,
it does not require matter for its propagation. Its maximal speed (2.998 108m/s)
occurs in a vacuum. In matter such as air, water, or glass, its speed is a function of
the transport characteristics of the medium. EM radiation travels in straight lines;
however, its trajectory can be altered by interaction with matter. The interaction of
EM radiation can occur by scattering (change in trajectory), absorption (removal of
the radiation), or, at very higher energies, transformation into particulate radiation
(energy to mass conversion).
EM radiation is commonly characterized by wavelength (l), frequency (n), and
energy per photon (E). EM radiation over a wide range of wavelengths, frequencies,
and energy per photon comprises the EM spectrum. For convenient reference, the
EM spectrum is divided into categories including the radio spectrum (that includes
transmissions from familiar technologies such as AM, FM, and TV broadcasting;
cellular and cordless telephone systems; as well as other wireless communications
technologies); infrared radiation (i.e., radiant heat); visible, ultraviolet (UV); and
x-ray and gamma-ray regions (Fig. 2-1).
Several forms of EM radiation are used in diagnostic imaging. Gamma rays,
emitted by the nuclei of radioactive atoms, are used to image the distributions of
radiopharmaceuticals. X-rays, produced outside the nuclei of atoms, are used in radiography, fluoroscopy, and computed tomography. Visible light is produced when
x-rays or gamma rays interact with various scintillators in the detectors used in several imaging modalities and is also used to display images. Radiofrequency EM radiation, near the FM frequency region, is used as the excitation and reception signals for
magnetic resonance imaging.
Wave-Particle Duality
There are two equally correct ways of describing EM radiationas waves and as
discrete particle-like packets or quanta of energy called photons. A central tenet of
quantum mechanics is that all particles exhibit wave-like properties and all waves
exhibit particle-like properties. Waveparticle duality addresses the inadequacy
of classical Newtonian mechanics in fully describing the behavior of atomic and
sub-atomic-scale objects.
18
Size
reference
105
101
108
Radio Spectrum
107
102
103
1 mm
MRI (1.5T)
63.86 MHz
SmallMeter
0.9 & 2.45 GHz
1011
107
1016
Infrared
Visible wavelengths (nm)
Terahertz
mm wave
sub-mm
Screening
0.2-4.0 THz
Night Vision
10-0.7 m
Remotes
850 nm
Dental Curing
200-350nm
Suntan
400-290nm
Visible Light
700-400nm
Fiber telecom
0.7-1.4 m
1A
1010
1011 1012
1019
105
1 ZHz
1020 1021
106
Bone Scan
140 keV
Medical X-ray
80 kev
Baggage Screen
160 keV
Cosmic
Gamma Rays
>10Bev
PET
511 kev
Hard X-ray
X-rays & Gamma Rays
1018
1 EHz
104
Soft X-ray
1017
103
Ionization DNA Damage
102
Ultraviolet
1015
10
1 PHz
1014
1 nm
109
atomic
nucleus
1 pm
Ionizing Radiation
108
DNA
atom
Chapter 2 Radiation and the Atom
FIGURE 2-1 The EM spectrum.
Microwave Oven
2.4 GHz
Wireless Data
~2.4 GHz
101
1013
106
1 m
Ex & Phcm
105
1012
102
104
bacteria
viruses
1 THz
103
Mobile Phones RADAR

900MHz-2.4GHz 1-100 GHz
TV Broadcast
54-700 MHz
FM radio
88-108 MHz
1010
104
Heating
Microwave
109
105
1 GHz
106
106
107
1 MHz
108
Broadcast and Wireless
109
Non-ionizing Radiation
102
Induced Currents
AM radio
600kHz-1.6MHz
frequency
(Hz)
electron volt
(eV)
Biological
Effect
102
optics
103
700
wavelength
l (m)
paper
thickness
625
baseball
1 ft
1 cm
blood
cells
575
football field
paperclip
thickness
540
mans height
470
Bands
Sources and Uses of

Frequency Bands
ELECTROMAGNETIC RADIATION SPECTRUM
electronics
440

19
20
Light Enters
Photocathode
Relative Size
FIGURE 2-2 Wave- and particle-like properties of light. A. Colors on the CD are produced as light waves
interact with the periodic structure of the tracks on a CD. This diffraction grating effect of the CD tracks splits
and diffracts light into several beams of different frequencies (color) traveling in different directions. B. The
imaging chain in nuclear medicine begins when gamma rays, which are not intercepted by the lead collimator, interact with the NaI crystal of a gamma camera (not shown) producing light photons. The NaI crystal is
optically coupled to the surface of a number of PMTs like the one shown above. Just below the glass surface
of the PMT, light photons strike the photocathode, ejecting electrons in a classical billiard ball (particle-like)
fashion. The ejected electrons are subsequently accelerated and amplified in the PMT, thus increasing gain of
the signals used to localize the gamma ray interactions. Further details of nuclear medicine imaging systems
are provided in Chapters 18 and 19.
Wave characteristics are more apparent when EM radiation interacts with objects
of similar dimensions as the photons wavelength. For example, light is separated into
colors by the diffraction grating effect of the tracks on a compact disc (CD) where the
track separation distance is of the same order of magnitude as the wavelength of the
visible light, Figure 2-2A.
Particle characteristics of EM radiation, on the other hand, are more evident
when an objects dimensions are much smaller than the photons wavelength. For
example, UV and visible light photons exhibit particle-like behavior during the
detection and localization of gamma rays in a nuclear medicine gamma camera.
Light photons, produced by the interaction of gamma rays with the NaI crystal
of a gamma camera, interact with and eject electrons from atoms in the photocathode of a photomultiplier tubes (PMTs) (Fig. 2-2B). The PMTs are optically
coupled to the crystal, thereby producing electrical signals for image formation
(discussed in greater detail in Chapter18). The particle-like behavior of x-rays
is exemplified by the classical billiard-ball type of collision between an x-ray
photon and an orbital electron during a Compton scattering event. Similarly the
x-ray photons energy is completely absorbed by, and results in the ejection of, an
orbital electron (a photoelectron), in the photoelectric effect. Each of these interactions is important to medical imaging and will be discussed in greater detail in
Chapter 3. Prior to the development of quantum mechanics, the classical wave
description of EM radiation could not explain the observation that the kinetic
energies of the photoelectrons were dependent on the energy (or wavelength) of
the incident radiation, rather than the intensity or quantity of incident photons.
Albert Einstein received the Nobel Prize in 1921 for his explanation of the photoelectric effect.
21
Wave Characteristics
Any wave (EM or mechanical, such as sound) can be characterized by their amplitude (maximal height), wavelength (l), frequency (n), and period (t). The amplitude
is the intensity of the wave. The wavelength is the distance between any two identical points on adjacent cycles. The time required to complete one cycle of a wave
(i.e., one l) is the period. The number of periods that occur per second is the frequency (1/t). Phase is the temporal shift of one wave with respect to the other. Some
of these quantities are depicted in Figure 2-3. The speed (c), wavelength, and frequency of any wave are related by
c v
[2-1]
Because the speed of EM radiation is constant in a given medium, its frequency

and wavelength are inversely proportional. Wavelengths of x-rays and gamma rays
are typically measured in fractions of nanometers (nm), where 1 nm 109 m.
Frequency is expressed in hertz (Hz), where 1 Hz 1 cycle/s 1 s1.
EM radiation propagates as a pair of oscillating and mutually reinforcing electric
and magnetic fields that are orthogonal (perpendicular) to one another and to the
direction of propagation, as shown in Figure 2-4.
Problem: Find the frequency of blue light with a wavelength of 400 nm in a
vacuum.
Solution: From Equation 2-1,
v
c (3 108 m s 1)(109 nm m1 )
7.5 1014 s 1 7.5 1014 Hz
400 nm
Particle Characteristics
The discrete (particle-like) packets (or quanta) of EM energy are called photons. The
energy of a photon is given by

E hv
hc

[2-2]
where h (Plancks constant) 6.6261034 J-s 4.1361018 keV-s. When E is

expressed in keV and l in nanometers (nm),
E (keV)
1.24

(nm)
[2-3]
= wavelength
Electric field
3/4 cycle (270) out of phase

1
Amplitude
Distance
2
Magnetic
field
Direction of waveform travel
FIGURE 2-3 Characterization of waves.
22
Electric field
= wavelength
Magnetic
field
Direction of waveform travel
FIGURE 2-4 Electric and magnetic field components of EM radiation.
The energies of photons are commonly expressed in electron volts (eV). One
electron volt is defined as the energy acquired by an electron as it traverses an electrical potential difference (voltage) of one volt in a vacuum. Multiples of the eV common to medical imaging are the keV (1,000 eV) and the MeV (1,000,000 eV).
Ionizing Radiation
An atom or molecule that has lost or gained one or more electrons has a net electrical
charge and is called an ion (e.g., sodium ion or Na). Some but not all electromagnetic and particulate radiations can cause ionization. In general, photons of higher
frequency than the far UV region of the spectrum (i.e., wavelengths greater than
200 nm) have sufficient energy per photon to remove bound electrons from atomic
shells, thereby producing ionized atoms and molecules. Radiation in this portion of
the spectrum (e.g., x-rays and gamma rays) is called ionizing radiation. EM radiation
with photon energies in and below the UV region (e.g., visible, infrared, terahertz,
microwave and radio waves) is called nonionizing radiation.
The threshold energy for ionization depends on the type and state of matter.
The minimum energies necessary to remove an electron (referred to as the ionization
energy) from calcium (Ca), glucose (C6H12O6), and liquid water (H2O) are 6.1, 8.8,
and 11.2 eV respectively. As water is the most abundant (thus most likely) molecular
target for radiation interaction in the body, a practical radiobiological demarcation
between ionizing and nonionizing EM radiation is approximately 11 eV. While 11 eV
is the lowest photon energy capable of producing ionization in water, in a random set
of ionization events evoked in a medium by ionization radiation, the average energy
expended per ion pair (W) is larger than the minimum ionization energy. For water
and tissue equivalent gas, W is about 30 eV. Particulate radiations such high speed
electrons and alpha particles (discussed below) can also cause ionization. Particulate
and EM ionizing radiation interactions are discussed in more detail in Chapter 3.
Particulate Radiation
The physical properties of the most important particulate radiations associated with medical imaging are listed in Table 2-1. Protons are found in the nuclei of all atoms. A proton
has a positive electrical charge and is identical to the nucleus of a hydrogen-1 atom. An
atomic orbital electron has a negative electrical charge, equal in magnitude to that of a
proton, and is approximately 1/1,800 the mass of a proton. Electrons emitted by the
nuclei of radioactive atoms are referred to as beta particles. Except for their nuclear origin,
negatively charged beta-minus particles (b2), or negatrons, are indistinguishable from
ordinary orbital electrons. However, there are also positively charged electrons, referred to
as beta-plus particles (b), or positrons; they are a form of antimatter that ultimately combines with matter in a unique transformation in which all of their mass is instantaneously
23
TABLE 2-1 PROPERTIES OF PARTICULATE RADIATION

PARTICLE
SYMBOL
Elementary
CHARGE
Rest
MASS (amu)
ENERGY
EQUIVALENT (mev)
Alpha
a, 4He22
12
4.00154
3,727
Proton
p, H
11
1.007276
938
Electron
0.000549
0.511
Negatron (beta minus)
0.000549
0.511
Positron (beta plus)
0.000549
0.511
Neutron
n0
1.008665
940
amu, atomic mass unit , defined as 1/12th the mass of a carbon-12 atom. Elementary charge is a unit of electric
charge where 1 is equal in magnitude to the charge of an electron.
converted to an equivalent amount of energy in the form of high-energy gamma rays.

This mass-energy conversion and its relevance to medical imaging are discussed briefly
below and in more detail in Chapters 3 and 15. Unless otherwise specified, common
usage of the term beta particle refers to b, whereas b particles are usually referred to as
positrons. A neutron is an uncharged nuclear particle that has a mass slightly greater than
that of a proton. Neutrons are released by nuclear fission and are used for radionuclide
production (Chapter 16). An alpha particle (a) consists of two protons and two neutrons;
it thus has a 2 charge and is identical to the nucleus of a helium atom (4He2). Alpha
particles are emitted by many high atomic number radioactive elements, such as uranium, thorium, and radium. Following emission, the a particle eventually acquires two
electrons from the surrounding medium and becomes an uncharged helium atom (4He).
Whereas alpha particles emitted outside the body are harmless, alpha particles emitted
inside the body cause more extensive cellular damage per unit energy deposited in tissue than any type of radiation used in medical imaging. The emission of alpha particles
during radioactive decay is discussed in Chapter 15, and the radiobiological aspects of
internally deposited alpha particles are discussed in Chapter 20.
Mass-Energy Equivalence
Within months of Einsteins ground breaking work uniting the concepts of space
and time in his special theory of relativity, he theorized that mass and energy were
also two aspects of the same entity and in fact were interchangeable. In any reaction,
the sum of mass and energy must be conserved. In classical physics, there are two
separate conservation laws, one for mass and one for energy. While these separate
conservation laws provided satisfactory explanations of the behavior of objects moving at relatively low speeds, they fail to explain some nuclear processes in which
particles approach the speed of light. For example, the production of the pairs of 511
keV annihilation photons used in position emission tomography (PET) could not
be explained if not for Einsteins insight that neither mass nor energy is necessarily
conserved separately, but can be transformed, one into the other, and it is only the
total mass-energy that is always conserved. The relationship between the mass and
the energy is expressed in one of the most famous equations in science:
E mc 2
[2-4]
where E (in joules -symbol J where 1 kg m2 s22 5 1 J) represents the energy equivalent
to mass m at rest and c is the speed of light in a vacuum (2.998 108 m/s). For example,
the energy equivalent of an electron with a rest mass (m) of 9.109 1031 kg is
24
E mc 2
E (9.109 1031 kg) (2.998 108 m/s)2
E 8.187 1014 J (8.187 1014 J) (1 MeV/1.602 1013 J)
0.511 Mev 511 keV
The conversion between mass and energy occurs in other phenomena discussed
later in this book including pair production, radioactive decay and annihilation radiation (used in positron emission tomography), (Chapters 3, 15 and 19
respectively).
A common unit of mass used in atomic and nuclear physics is the atomic mass
unit (amu), defined as 1/12th of the mass of an atom of 12C. One amu is equivalent to
931.5 MeV of energy.
2.2
Structure of the Atom

The atom is the smallest division of an element in which the chemical identity of
the element is maintained. The atom is composed of an extremely dense positively
charged nucleus, containing protons and neutrons, and an extranuclear cloud of
light negatively charged electrons. In its nonionized state, an atom is electrically neutral because the number of protons equals the number of electrons. The radius of an
atom is approximately 1010 m, whereas that of the nucleus is only about 1014m.
Thus, the atom is largely unoccupied space, in which the volume of the nucleus is
only 1012 (a millionth of a millionth) the volume of the atom. If the empty space
in an atom could be removed, a cubic centimeter of protons would have a mass of
approximately 4 million metric tons!
Electron Orbits and Electron Binding Energy

In the Bohr model of the atom (Niels Bohr 1913), electrons orbit around a dense,
positively charged nucleus at fixed distances (Bohr radii). Bohr combined the classical Newtonian laws of motion and Coulombs law of electrostatic attraction with
quantum theory. In this model of the atom, each electron occupies a discrete energy
state in a given electron shell. These electron shells are assigned the letters K L,
M, N,, with K denoting the innermost shell, in which the electrons have the lowest energies. The shells are also assigned the quantum numbers 1, 2, 3, 4,, with the
quantum number 1 designating the K shell. Each shell can contain a maximum number of electrons given by (2n2), where n is the quantum number of the shell. Thus,
the K shell (n 1) can only hold 2 electrons, the L shell (n 2) can hold 2(2)2 or
8 electrons, and so on, as shown in Figure 2-5. The outer electron shell of an atom,
the valence shell, determines the chemical properties of the element. Advances in
atomic physics and quantum mechanics led to refinements of the Bohr model.
According to contemporary views on atomic structure, the location of an orbital
electron is more properly described in terms of the probability of its occupying a
given location within the atom with both wave and particle properties. At any given
moment, there is even a probability, albeit very low, that an electron can be within the
atoms nucleus. However, the highest probabilities are associated with Bohrs original
atomic radii.
The energy required to remove an orbital electron completely from the atom is
called its orbital binding energy. Thus, for radiation to be ionizing, the energy transferred
25
O P Q
Nucleus
Quantum #
6 7
Maximum
electron
capacity
18
32
50 72 98
FIGURE 2-5 Electron shell designations and orbital filling rules.
to the electron must equal or exceed its binding energy. Due to the closer proximity
of the electrons to the positively charged nucleus, the binding energy of the K-shell
is greater than that of outer shells. For a particular electron shell, binding energy
also increases with the number of protons in the nucleus (i.e., atomic number). In
Figure 2-6, electron binding energies are compared for hydrogen (Z 1) and tungsten
(Z 74). A K shell electron of tungsten with 74 protons in the nucleus is much more
Valence e
Zero
Valence e
Zero
1.5
3.4
L series
Energy (eV)
Energy (eV)
N
2,500
11,000
K series
K series
13.5
Hydrogen Z = 1
L series
69,500
Tungsten Z = 74
FIGURE 2-6 Energy-level diagrams for hydrogen and tungsten. The energy necessary to separate electrons
in particular orbits from the atom (not drawn to scale) increases with Z and decreases with distance from
the nucleus. Note that zero energy represents the point at which the electron is experiencing essentially no
Coulomb attractive force from the protons in the nucleus (often referred to as a free electron). For a bound
electron to reach that state, energy has to be absorbed. Thus, the energy states of the electrons within the
atom must be below zero and are thus represented as negative numbers. The vertical lines represent various
transitions (e.g., K and L series) of the electrons from one energy level to another.
26
tightly bound (~69,500 eV) than the K shell electron of hydrogen orbiting a nucleus
with a single proton (~13.5 eV). The energy required to move an electron from the
innermost electron orbit (K shell) to the next orbit (L shell) is the difference between
the binding energies of the two orbits (i.e., EbK EbL equals the transition energy).
Hydrogen:
13.5 eV 3.4 eV 10.1 eV
Tungsten:
69,500 eV 11,000 eV 58,500 eV (58.5 keV)
Radiation from Electron Transitions

When an electron is removed from its shell by an x-ray or gamma ray photon or a
charged particle interaction, a vacancy is created in that shell. This vacancy is usually
filled by an electron from an outer shell, leaving a vacancy in the outer shell that in
turn may be filled by an electron transition from a more distant shell. This series of
transitions is called an electron cascade. The energy released by each transition is equal
to the difference in binding energy between the original and final shells of the electron.
This energy may be released by the atom as characteristic x-rays or Auger electrons.
Characteristic x-rays
Electron transitions between atomic shells can result in the emission of radiation
in the visible, UV, and x-ray portions of the EM spectrum. The energy of this radiation is characteristic of each atom, since the electron binding energies depend on Z.
Emissions from transitions exceeding 100 eV are called characteristic or fluorescent
x-rays. Characteristic x-rays are named according to the orbital in which the vacancy
occurred. For example, the radiation resulting from a vacancy in the K shell is called
a K-characteristic x-ray, and the radiation resulting from a vacancy in the L shell is
called an L characteristic x-ray. If the vacancy in one shell is filled by the adjacent
shell, it is identified by a subscript alpha (e.g., L K transition Ka, M L transition La). If the electron vacancy is filled from a nonadjacent shell, the subscript beta
is used (e.g., M K transition Kb). The energy of the characteristic x-ray (Ex-ray) is
the difference between the electron binding energies (Eb) of the respective shells:

E x-ray E b vacant shell E b transition shell
[2-5]
Thus, as illustrated in Figure 2-7A, an M to K shell transition in tungsten would

produce a Kb characteristic x-ray of
E (K ) E bK E bM
E (K ) 69.5 keV 2.5 keV 67 keV
Auger Electrons and Fluorescent Yield

An electron cascade does not always result in the production of a characteristic xray
or x-rays. A competing process that predominates in low Z elements is Auger electron
emission. In this case, the energy released is transferred to an orbital electron, typically in the same shell as the cascading electron (Fig. 2-7B). The ejected Auger electron possesses kinetic energy equal to the difference between the transition energy
and the binding energy of the ejected electron.

2.5 keV
11 keV
Cascading
electron
2.5 keV
67 keV K
Characteristic 11 keV
Cascading
X-ray
electron
69.5 keV
Vacant
69.5 keV
Vacant
64.5 keV
Auger
electron
Energy
transfer
27
FIGURE 2-7 De-excitation of a tungsten atom. An electron transition filling a vacancy in an orbit closer to
the nucleus will be accompanied by either the emission of characteristic radiation (A) or the emission of an
Auger electron (B).
The probability that the electron transition will result in the emission of a characteristic x-ray is called the fluorescent yield (). Thus, 1 is the probability that
the transition will result in the ejection of an Auger electron. Auger emission predominates in low Z elements and in electron transitions of the outer shells of heavy
elements. The K-shell fluorescent yield is essentially zero (1%) for elements Z ,10
(i.e., the elements comprising the majority of soft tissue), about 15% for calcium
(Z20), about 65% for iodine (Z 53), and approaches 80% for Z . 60.
The Atomic Nucleus

Composition of the Nucleus
The nucleus is composed of protons and neutrons, known collectively as nucleons.
The number of protons in the nucleus is the atomic number (Z), and the total number
of protons and neutrons within the nucleus is the mass number (A). It is important
not to confuse the mass number with the atomic mass, which is the actual mass of the
atom. For example, the mass number of oxygen-16 is 16 (8 protons and 8 neutrons),
whereas its atomic mass is 15.9994 amu. The notation specifying an atom with the
chemical symbol X is AZ X N , where N is the number of neutrons in the nucleus. In
this notation, Z and X are redundant because the chemical symbol identifies the element and thus the number of protons. For example, the symbols H, He, and Li refer
to atoms with Z 1, 2, and 3, respectively. The number of neutrons is calculated as
131
N A Z. For example, 131
I or as I-131. The charge
53 I 78 is usually written as
on an atom is indicated by a superscript to the right of the chemical symbol. For
example, Ca2 indicates that the calcium atom has lost two electrons and thus has a
net charge of 2.
Nuclear Forces and Energy Levels

There are two main forces that act in opposite directions on particles in the
nucleus. The coulombic force between the protons is repulsive and is countered by the attractive force resulting from the exchange of gluons (subnuclear
particles) among all nucleons. The exchange forces, also called the strong force,
hold the nucleus together but operate only over very short (nuclear) distances
(,1014 m).
The nucleus has energy levels that are analogous to orbital electron shells,
although often much higher in energy. The lowest energy state is called the ground
28
TABLE 2-2NUCLEAR FAMILIES: ISOTOPES, ISOBARS, ISOTONES,
AND ISOMERS
FAMILY
NUCLIDES WITH SAME
EXAMPLE
Isotopes
Atomic number (Z)
I-131 and I-125: Z 5 53
Isobars
Mass number (A)
Mo-99 and Tc-99: A 5 99
Isotones
Number of neutrons (AZ)
53
Isomers
Atomic and mass numbers

but different energy states
in the nucleus
Tc-99m and Tc-99: Z 5 43

A 5 99
Energy of Tc-99m . Tc-99: E 5 142 keV
I-131: 131 53 5 78
Xe-132: 132 54 5 78
54
Note: See text for description of the italicized letters in the nuclear family terms.
state of an atomic nucleus. Nuclei with energy in excess of the ground state are said
to be in an excited state. The average lifetimes of excited states range from 1016 s to
more than 100 y. Excited states that exist longer than 1012 s are referred to as metastable or isomeric states. Metastable states with longer lifetimes are denoted by the
letter m after the mass number of the atom (e.g., Tc-99m).
Classification of Nuclides
Species of atoms characterized by the number of protons and neutrons and the
energy content of the atomic nuclei are called nuclides. Isotopes, isobars, isotones,
and isomers are families of nuclides that share specific properties (Table 2-2). An
easy way to remember these relationships is to associate the p in isotopes with the
same number of protons, the a in isobars with the same atomic mass number, the n
in isotones with the same number of neutrons, and the e in isomer with the different
nuclear energy states.
Nuclear Stability and Radioactivity

Only certain combinations of neutrons and protons in the nucleus are stable; the others are radioactive. On a plot of Z versus N, these stable nuclides fall along a line of
stability for which the N/Z ratio is approximately 1 for low Z nuclides and approximately 1.5 for high Z nuclides, as shown in Figure 2-8. A higher neutron-to-proton
ratio is required in heavy elements to offset the Coulomb repulsive forces between
protons. Only four nuclides with odd numbers of neutrons and odd numbers of
protons are stable, whereas many more nuclides with even numbers of neutrons and
even numbers of protons are stable. The number of stable nuclides identified for
different combinations of neutrons and protons is shown in Table 2-3. Nuclides with
an odd number of nucleons are capable of producing a nuclear magnetic resonance
signal, as described in Chapter 12.
While atoms with unstable combinations of neutrons and protons exist, over
time they will transform to nuclei that are stable. Two kinds of instability are neutron
excess and neutron deficiency (i.e., proton excess). Such nuclei have excess internal
energy compared with a stable arrangement of neutrons and protons. They achieve
stability by the conversion of a neutron to a proton or vice versa, and these events
are accompanied by the emission of energy. The energy emissions include particulate
0
26
Lines of isobars: same number of nucleons (A)
0
28 0
27
180
Lines of isotopes: same number of protons (Z)

Lines of isotones: same number of neutrons (N)
0
21
N/Z~1.5
(e.g. 98Cf252)
0 0
20 19
120
100 yr
14
0
1 yr
106 sec
80
11
N/Z = 1
Neutron number N = A-Z
0
18
0
17
104 yr
13
12
108 yr
106 yr
0
16
0
15
100
1014 yr
1010 yr
0
23
0
22
140
stable
1012 yr
0
25
0
24
160
29
104 sec
1
90 00
100 sec
60
80
60
40
1 sec
70
102 sec
50
40
104 sec
30
20
106 sec
20
Z = 83 (Bismuth)
N/Z = 1
(e.g. 8O16)
10
20
30
No stable nuclides
beyond this point
40
50
60
70
Proton number Z
80
90
100
110 120
108 sec
most unstable
no nuclide
FIGURE 2-8 A plot of the nuclides where the number of protons (i.e., atomic number or Z) and neutrons
of each nuclide is shown on the x- and y-axes, respectively. The outlined area containing all the black and
colored squares represents the range of known nuclides. The stable nuclides are indicated by small black
squares, whereas the colored squares represent radioactive (i.e., unstable) nuclides or radionuclides. The stable
nuclides form the so-called line of stability in which the neutron-to-proton ratio is approximately 1 for low Z
nuclides and increases to approximately 1.5 for high Z nuclides. The color code indicates the physical half-life
of each of the radionuclides. Note that all nuclides with Z > 83 (bismuth) are radioactive and that, in general,
the further the radionuclide is from the line of stability, the more unstable it is and the shorter the half-life
it has. Radionuclides to the left of the line of stability are neutron rich and are likely to undergo beta-minus
decay, while radionuclides to the right of the line of stability are neutron poor and thus often decay by positron
emission or electron capture. Extremely unstable radionuclides and those with high Z often decay by alpha
particle emission.
30
TABLE 2-3DISTRIBUTION OF STABLE NUCLIDES AS A FUNCTION
OF NEUTRON AND PROTON NUMBER

NUMBER OF PROTONS (Z)
NUMBER OF NEUTRONS (N)
NUMBER OF STABLE NUCLIDES
Even
Even
165
Even
Odd
57 (NMR signal)
Odd
Even
53 (NMR signal)
Odd
Odd
4 (NMR signal)
Total
279
NMR, nuclear magnetic resonance.
and EM radiations. Nuclides that decay (i.e., transform) to more stable nuclei are
said to be radioactive, and the transformation process itself is called radioactive decay
(radioactive disintegration). There are several types of radioactive decay, and these
are discussed in detail in Chapter 15. A nucleus may undergo several decays before
a stable configuration is achieved. These decay chains are often found in nature.
For example, the decay of uranium 238 (U-238) is followed by 13 successive decays
before the stable nuclide, lead 206 (Pb-206), is formed. The radionuclide at the
beginning of a particular decay sequence is referred to as the parent, and the nuclide
produced by the decay of the parent is referred to as the daughter. The daughter may
be either stable or radioactive.
Gamma Rays
Radioactive decay often results in the formation of a daughter nucleus in an
excited state. The EM radiation emitted from the nucleus as the excited state
transitions to a lower (more stable) energy state is called a gamma ray. This energy
transition is analogous to the emission of characteristic x-rays following electron
transition. However, gamma rays (by definition), emanate from the nucleus. Because
the spacing of the energy states within the nucleus is usually considerably larger
than those of atomic orbital electron energy states, electron transitions, gamma
rays are often much more energetic than characteristic x-rays. When this nuclear
de-excitation process takes place in an isomer (e.g., Tc-99m), it is called isomeric
transition (discussed in Chapter 15). In isomeric transition, the nuclear energy state
is reduced with no change in A or Z.
Internal Conversion Electrons

Nuclear de-excitation does not always result in the emission of a gamma ray. An
alternative form of de-excitation is internal conversion, in which the de-excitation
energy is completely transferred to an orbital (typically K, L, or M shell) electron.
The conversion electron is ejected from the atom, with a kinetic energy equal to that
of the gamma ray less the electron binding energy. The vacancy produced by the
ejection of the conversion electron will be filled by an electron cascade and associated characteristic x-rays and Auger electrons as described previously. The internal
conversion energy transfer process is analogous to the emission of an auger electron
in lieu of characteristic x-ray energy emission. However, the kinetic energy of the
internal conversion electron is often much greater than that of Auger electrons due
to the greater energy associated with most gamma ray emissions compared to characteristic x-ray energies.
31
Nuclear Binding Energy and Mass Defect

The energy required to separate an atom into its constituent parts is the atomic
binding energy. It is the sum of the orbital electron binding energy and the nuclear
binding energy. The nuclear binding energy is the energy necessary to disassociate a
nucleus into its constituent parts and is the result of the strong forces acting between
nucleons. Compared with the nuclear binding energy, the orbital electron binding
energy is negligible. When two subatomic particles approach each other under the
influence of this strong nuclear force, their total energy decreases and the lost energy
is emitted in the form of radiation. Thus, the total energy of the bound particles is
less than that of the separated free particles.
The binding energy can be calculated by subtracting the mass of the atom from
the total mass of its constituent protons, neutrons, and electrons; this mass difference
is called the mass defect. For example, the mass of an N-14 atom, which is composed
of 7 electrons, 7 protons, and 7 neutrons, is 14.00307 amu. The total mass of its
constituent particles in the unbound state is
mass of 7 protons 7(1.007276 amu)
mass of 7 neutrons 7(1.008665 amu)
mass of 7 electrons 7(0.000549 amu)
mass of component particles of N-14
7.050932 amu
7.060655 amu
0.003843 amu
14.11543 amu
Thus, the mass defect of the N-14 atom, the difference between the mass of its
constituent particles and its atomic mass, is 14.11543 amu 14.003074 amu
0.112356 amu. According to the formula for mass energy equivalence (Equation 2-4),
this mass defect is equal to (0.112356 amu) (931.5 MeV/amu) 104.7 MeV.
An extremely important observation is made by carrying the binding energy calculation a bit further. The total binding energy of the nucleus may be divided by the
mass number A to obtain the average binding energy per nucleon. Figure 2-9 shows
the average binding energy per nucleon for stable nuclides as a function of mass
number. The fact that this curve reaches its maximum near the middle elements and
decreases at either end predicts that large quantities of energy can be released from a
small amount of matter. The two processes by which this can occur are called nuclear
fission and nuclear fusion.
Nuclear Fission and Fusion

During nuclear fission, a nucleus with a large atomic mass splits into two usually
unequal parts called fission fragments, each with an average binding energy per
nucleon greater than that of the original nucleus. In this reaction, the total nuclear
binding energy increases. The change in the nuclear binding energy is released as
EM and particulate radiation and as kinetic energy of the fission fragments. Fission is
typically accompanied by the release of several neutrons. For example, the absorption
of a single neutron by the nucleus of a U-235 atom can result in the instantaneous fission of U-236 into two fission fragments (e.g., Sn-131 and Mo-102), and two or three
energetic neutrons. This reaction results in a mass defect equivalent to approximately
200 MeV, which is released as kinetic energy of the fission fragments (~165 MeV);
neutrons (~5 MeV); prompt (instantaneous) gamma radiation (~7 MeV) and radiation
from the decay of the fission products (~23 MeV). The probability of fission increases
with the neutron flux, which is the number of neutrons per cm2/s. Fission is used in
nuclear powered electrical generating plants and in the design of atom bombs. The
use of nuclear fission for radionuclide production is discussed in Chapter 16.
32
Binding energy per nucleon (EB /A), MeV
9
16O
12C
8
7
4He
6
5
6Li
4
3
F
U
S
I
O
N
56Fe
98Mo 119
Sn
H-2
H-3
208Pb
235U 238
U
U-235
Neutron
+14.1 MeV
U-236
He-4
+3.5 MeV
3H
3He
181Ta
Fission Fragments
+
~165 MeV
Kr-82
Ba-141
F
I
S
S
I
O
N
Prompt gamma
~7 MeV
3n+~5Mev
2H
0
0
50
100
150
Mass number A
200
250
FIGURE 2-9 Average binding energy per nucleon.
Energy is also released from the fusion (combining) of light atomic nuclei. For
example, the fusion of deuterium (H-2) and tritium (He-3) nuclei results in the production of helium-4 and a neutron. As can be seen in Figure 2-9, the helium-4 atom
has a much higher average binding energy per nucleon than either tritium (H-3) or
deuterium (H-2). The energy associated with the mass defect of this reaction is about
17.6 MeV. The nuclei involved in fusion reactions require exceptionally high kinetic
energies in order to overcome Coulomb repulsive forces. Fusion of hydrogen nuclei in
stars, where extreme gravitational forces result in extraordinarily high temperatures,
is the first step in subsequent self-sustaining fusion reactions that produce helium-4.
The H-bomb is a fusion device that uses the detonation of an atom bomb (a fission
device) to generate the temperature and pressure necessary for fusion. H-bombs are
also referred to as thermonuclear weapons.
SUGGESTED READING
Cherry, R., Simon et.al. Physics in Nuclear Medicine. 4th ed., Philadelphia: Saunders, 2011.
Evans RD. The atomic nucleus. Malabar, FL: Robert E. Krieger, 1982.
Turner, J. E. Atoms, radiation, and radiation protection, 3rd ed. Weinheim, Germany: Wiley-VCH Verlag
GmbH & Co. KGaA 2007.
Chapter
Interaction of Radiation
with Matter
3.1 Particle Interactions
Particles of ionizing radiation include charged particles, such as alpha particles (12),
protons (p1), beta particles (2), positrons (1), energetic extranuclear electrons (e2)
and uncharged particles, such as neutrons. The behavior of heavy charged particles
(e.g., alpha particles and protons) is different from that of lighter charged particles
such as electrons and positrons.
Excitation, Ionization, and Radiative Losses

Energetic charged particles interact with matter by electrical (i.e., coulombic) forces
and lose kinetic energy via excitation, ionization, and radiative losses. Excitation and ionization occur when charged particles lose energy by interacting with orbital electrons
in the medium. These interactional, or collisional, losses refer to the coulombic forces
exerted on charged particles when they pass in proximity to the electric field generated
by the atoms electrons and protons. Excitation is the transfer of some of the incident
particles' energy to electrons in the absorbing material, promoting them to electron
orbits farther from the nucleus (i.e., higher energy level). In excitation, the energy
transferred to an electron does not exceed its binding energy. Following excitation, the
electron will return to a lower energy level, with the emission of the excitation energy
in the form of electromagnetic radiation or Auger electrons. This process is referred to
as de-excitation (Fig. 3-1A). If the transferred energy exceeds the binding energy of the
electron, ionization occurs, whereby the electron is ejected from the atom (Fig. 3-1B).
The result of ionization is an ion pair consisting of the ejected electron and the positively
charged atom. Sometimes, the ejected electrons possess sufficient energy to produce
further ionizations called secondary ionization. These electrons are called delta rays.
Approximately 70% of the energy deposition of energetic electrons in soft tissue
occurs via ionization. However, as electron energy decreases the probability of energy
loss via excitation increases. For very low energy electron (,40 eV) the probabilities
of excitation and ionization are equal and with further reductions in electron energy
the probability of ionization rapidly diminishes becoming zero (in tissue) below the
first ionization state of liquid water at approximately 11.2 eV. So, while the smallest
binding energies for electrons in carbon, nitrogen, and oxygen are less than 10 eV, the
average energy deposited per ion pair produced in air (mostly nitrogen and oxygen) and
soft tissue (mostly hydrogen, carbon, and oxygen) are approximately 34 eV and 22 eV,
respectively. The energy difference is the result of the excitation process. Medical imaging with x-rays and gamma rays results in the production of energetic electrons by
mechanisms discussed later in this chapter. It should be appreciated that, owing to the
relatively modest amount of energy necessary to produce a secondary electron, each of
these energetic electrons will result in a abundance of secondary electrons as they deposit
33
34

+
L
EMR
+ + +
+ + + +
+ + +
+ + + +
+ +
+ + +
+
Ionization caused by
Coulombic repulsion
Secondary electron
delta rays
Incident ionizing
+ or particle
+ + +
+ + + +
Positive Ion
B
FIGURE 3-1 A. Excitation (left) and de-excitation (right) with the subsequent release of electromagnetic
radiation. B. Ionization and the production of delta rays.
their energy in tissue. For example, a 10 keV electron will result in the production of
over 450 secondary electrons, most with energies between 10 and 70 eV.
Specific Ionization
The average number of primary and secondary ion pairs produced per unit length
of the charged particles path is called the specific ionization, expressed in ion pairs
(IP)/mm. Specific ionization increases with the square of the electrical charge (Q) of
the particle and decreases with the square of the incident particle velocity (v); thus,
Q2
SI 2 . A larger charge produces a greater coulombic field; as the particle loses
v
Chapter 3 Interaction of Radiation with Matter
Ion pairs/mm
103
7
35
FIGURE 3-2 Specific ionization

(ion pairs/mm) as a function of distance from the end of range in air
for a 7.69-MeV alpha particle from
214
Po. Rapid increase in specific ionization reaches a maximum (Bragg
peak) and then drops off sharply
as the particle kinetic energy is
exhausted and the charged particle
is neutralized.
Bragg ionization peak
6
5
4
3
2
1
0
5
4
3
2
1
Distance from End of Range, cm Air
kinetic energy, it slows down, allowing the coulombic field to interact at a given
location for a longer period of time. The kinetic energies of alpha particles emitted
by naturally occurring radionuclides extend from a minimum of about 4.05 MeV
(Th-232) to a maximum of about 10.53 MeV (Po-212). The ranges of alpha particles
in matter are quite limited and, for the alpha particle energies mentioned above, their
ranges in air are 2.49 and 11.6 cm respectively. In tissue the alpha particle range is
reduced to less than the diameter of a dozen or so cells, (~30 to 130 mm). The specific ionization of an alpha particle can be as high as approximately 7,000 IP/mm in
air and about 10 million IP/mm in soft tissue. The specific ionization as a function of
the particles path is shown for a 7.69-MeV alpha particle from 214Po in air (Fig. 3-2).
As the alpha particle slows, the specific ionization increases to a maximum (called
the Bragg peak), beyond which it decreases rapidly as the alpha particle acquires
electrons and becomes electrically neutral, thus losing its capacity for further ionization. The large Bragg peak associated with heavy charged particles has applications in
radiation therapy. For example, several proton therapy centers have been built over
the past decade. By adjusting the kinetic energy of heavy charged particles, a large
radiation dose can be delivered at a particular depth and over a fairly narrow range
of tissue containing a lesion. On either side of the Bragg peak, the dose to tissue is
substantially lower. Heavy particle accelerators are used at some medical facilities to
provide this treatment in lieu of surgical excision or conventional radiation therapy.
Compared to heavy charged particles, the specific ionization of electrons is much
lower (in the range of 5 to 10 IP/mm of air).
Charged Particle Tracks

Another important distinction between heavy charged particles and electrons is their
paths in matter. Electrons follow tortuous paths in matter as the result of multiple
scattering events caused by coulombic deflections (repulsion and/or attraction). The
sparse tortuous ionization track of an electron is illustrated in Figure 3-3A. On the
other hand, the larger mass of a heavy charged particle results in a dense and usually linear ionization track (Fig. 3-3B). The path length of a particle is defined as
the distance the particle travels. The range of a particle is defined as the depth of
penetration of the particle in matter. As illustrated in Figure 3-3, the path length of
the electron almost always exceeds its range, whereas the typically straight ionization track of a heavy charged particle results in the path length and range being
nearly equal. Additional information on the pattern of energy deposition of charged
particles at the cellular level and their radiobiological significance is presented in
Chapter 20.
36

+
+ +
++
+ +
+ + + + + + ++++
Path > Range
Path = Range
FIGURE 3-3 A. Electron scattering results in the path length of the electron being greater than its range.
B. Heavily charged particles, like alpha particles, produce a dense, nearly linear ionization track, resulting in the
path and range being essentially equal.
Linear Energy Transfer

While specific ionization reflects all energy losses that occur before an ion pair is
produced, the linear energy transfer (LET) is a measure of the average amount of energy
deposited locally (near the incident particle track) in the absorber per unit path length.
LET is often expressed in units of keV or eV per m. The LET of a charged particle
is proportional to the square of the charge and inversely proportional to the particles
kinetic energy (i.e., LET Q2/Ek ). The LET of a particular type of radiation describes
the local energy deposition density, which can have a substantial impact on the biologic consequences of radiation exposure. In general, for a given absorbed dose, the
dense ionization tracks of high LET radiations (alpha particles, protons, etc.) deposit
their energy over a much shorter range and are much more damaging to cells than
the spare ionization pattern associated with low LET radiations. Low LET radiation
includes energetic electrons (e.g., b2 and b) and ionizing electromagnetic radiation
(gamma and x-rays, whose interactions set electrons into motion). By way of perspective the exposure of patients to diagnostic x-rays results in the production of energetic
electrons with an average LET of approximately 3 keV/mm in soft tissue, whereas the
average LET of 5-MeV alpha particles in soft tissue is approximately 100 keV/mm.
Despite their typically much higher initial kinetic energy, the range of high LET radiation is much less than that of low LET radiation. For example at the point where an
alpha particle and electron traversing tissue have the same kinetic energy, say 100 keV,
their range from that point will be 1.4 and 200m respectively.
Scattering
Scattering refers to an interaction that deflects a particle or photon from its original
trajectory. A scattering event in which the total kinetic energy of the colliding particles is unchanged is called elastic. Billiard ball collisions, for example, are elastic
(disregarding frictional losses). When scattering occurs with a loss of kinetic energy
(i.e., the total kinetic energy of the scattered particles is less than that of the particles
before the interaction), the interaction is said to be inelastic. For example, the process
of ionization can be considered an elastic interaction if the binding energy of the
electron is negligible compared to the kinetic energy of the incident electron (i.e.,
the kinetic energy of the ejected electron is equal to the kinetic energy lost by the
incident electron). If the binding energy that must be overcome to ionize the atom
is not insignificant compared to the kinetic energy of the incident electron (i.e., the
kinetic energy of the ejected electron is less than the kinetic energy lost by the incident electron), the process is said to be inelastic.
37
Radiative InteractionsBremsstrahlung
While most electron interactions with the atomic nuclei are elastic, electrons can
undergo inelastic interactions in which the path of the electron is deflected by the
positively charged nucleus, with a loss of kinetic energy. This energy is instantaneously
emitted as electromagnetic radiation (i.e., x-rays). Energy is conserved, as the energy
of the radiation is equal to the kinetic energy lost by the electron.
The radiation emission accompanying electron deceleration is called bremsstrahlung, a German word meaning braking radiation (Fig. 3-4). The deceleration of the
high-speed electrons in an x-ray tube produces the bremsstrahlung x-rays used in
diagnostic imaging.
Total bremsstrahlung emission per atom is proportional to Z2, where Z is the
atomic number of the absorber, and inversely proportional to the square of the mass
of the incident particle, that is, Z2/m2. Due to the strong influence of the particles
mass, bremsstrahlung production by heavier charged particles such as protons and
alpha particles will be less than one millionth of that produced by electrons.
The energy of a bremsstrahlung x-ray photon can be any value up to and including the entire kinetic energy of the deflected electron. Thus, when many electrons
undergo bremsstrahlung interactions, the result is a continuous spectrum of x-ray
energies. This radiative energy loss is responsible for the majority of the x-rays produced by x-ray tubes and is discussed in greater detail in Chapter 6.
Positron Annihilation
The fate of positrons (b1) is unlike that of negatively charged electrons (e2 and
b2) that ultimately become bound to atoms. As mentioned above, all energetic
electrons (positively and negatively charged) lose their kinetic energy by excitation, ionization, and radiative interactions. When a positron (a form of antimatter)
reaches the end of its range, it interacts with a negatively charged electron, resulting in the annihilation of the electron-positron pair and the complete conversion
of their rest mass to energy in the form of two oppositely directed 0.511-MeV
annihilation photons. This process occurs following radionuclide decay by positron
emission (see Chapter 15). Imaging of the distribution of positron-emitting radiopharmaceuticals in patients is accomplished by the detection of the annihilation
photon pairs during positron emission tomography (PET) (see Chapter 19). The
annihilation photons are not often emitted at exactly 180 degrees apart because
there is often a small amount of residual momentum in the positron when it interacts with the oppositely charged electron. This noncolinearity is not severe (~0.5
degree), and its blurring effect in the typical PET imaging system is not clinically
significant.
Bremsstrahlung
x-ray
Coulombic attraction
+
++ +
+ ++
+
+ + +
+ +
Nucleus
FIGURE 3-4 Radiative energy loss via

bremsstrahlung (braking radiation).
38
FIGURE 3-5 Schematic example of

collisional energy loss. An uncharged
particle (neutron) interacts with the
atomic nucleus of an atom resulting
in the ejection of a proton. This interaction results in transformation of
the atom into a new element with an
atomic number (Z) reduced by 1.
L
K
+
+
+
Neutron
Neutron Interactions
Unlike protons and electrons, neutrons, being uncharged particles, cannot cause
excitation and ionization via coulombic interactions with orbital electrons. They
can, however, interact with atomic nuclei, sometimes liberating charged particles
or nuclear fragments that can directly cause excitation and ionization (Fig. 3-5).
Neutrons often interact with atomic nuclei of light elements (e.g., H, C, O) by scattering in billiard balllike collisions, producing recoil nuclei that lose their energy
via excitation and ionization. In tissue, energetic neutrons interact primarily with the
hydrogen in water, producing recoil protons (hydrogen nuclei). Neutrons may also be
captured by atomic nuclei. Neutron capture results in a large energy release (typically
2 to 7 MeV) due to the large binding energy of the neutron. In some cases, one or
more neutrons are reemitted; in other cases, the neutron is retained, converting the
atom into a different isotope. For example, the capture of a neutron by a hydrogen
atom (1H) results in deuterium (2H) and the emission of a 2.22-MeV gamma ray,
reflecting the increase in the binding energy of the nucleus:
1
H 1 1n 2H 1 Gamma ray energy (E) 5 2.22 MeV
Some nuclides produced by neutron absorption are stable, and others are
r adioactive (i.e., unstable). As discussed in Chapter 2, neutron absorption in some
very heavy nuclides such as 235U can cause nuclear fission, producing very energetic
fission fragments, neutrons, and gamma rays. Neutron interactions important to the
production of radiopharmaceuticals are described in greater detail in Chapter 16.
3.2 X-ray and Gamma-Ray Interactions

When traversing matter, photons will penetrate without interaction, scatter, or be
absorbed. There are four major types of interactions of x-ray and gamma-ray photons
with matter, the first three of which play a role in diagnostic radiology and nuclear
medicine: (a) Rayleigh scattering, (b) Compton scattering, (c) photoelectric absorption, and (d) pair production.
Rayleigh Scattering
In Rayleigh scattering, the incident photon interacts with and excites the total atom,
as opposed to individual electrons as in Compton scattering or the photoelectric
effect (discussed later). This interaction occurs mainly with very low energy x-rays,
such as those used in mammography (15 to 30 keV). During the Rayleigh scattering event, the electric field of the incident photons electromagnetic wave expends
energy, causing all of the electrons in the scattering atom to oscillate in phase. The
atoms electron cloud immediately radiates this energy, emitting a photon of the same
Scattered
photon
Incident
photon
39
1 = 2
FIGURE 3-6 Rayleigh scattering. The diagram shows that the incident photon l1 interacts with an atom and
the scattered photon l2 is being emitted with the same wavelength and energy. Rayleigh scattered photons
are typically emitted in the forward direction fairly close to the trajectory of the incident photon. K, L, and M
are electron shells.
energy but in a slightly different direction (Fig. 3-6). In this interaction, electrons are
not ejected, and thus, ionization does not occur. In general, the average scattering
angle decreases as the x-ray energy increases. In medical imaging, detection of the
scattered x-ray will have a deleterious effect on image quality. However, this type of
interaction has a low probability of occurrence in the diagnostic energy range. In
soft tissue, Rayleigh scattering accounts for less than 5% of x-ray interactions above
70 keV and at most only accounts for about 10% of interactions at 30 keV. Rayleigh
interactions are also referred to as coherent or classical scattering.
Compton Scattering
Compton scattering (also called inelastic or nonclassical scattering) is the predominant
interaction of x-ray and gamma-ray photons in the diagnostic energy range with soft
tissue. In fact, Compton scattering not only predominates in the diagnostic energy range
above 26 keV in soft tissue but also continues to predominate well beyond diagnostic
energies to approximately 30 MeV. This interaction is most likely to occur between
photons and outer (valence)-shell electrons (Fig. 3-7). The electron is ejected from
the atom, and the scattered photon is emitted with some reduction in energy relative
to the incident photon. As with all types of interactions, both energy and momentum
must be conserved. Thus, the energy of the incident photon (E0) is equal to the sum of
the energy of the scattered photon (Esc) and the kinetic energy of the ejected electron
(Ee2), as shown in Equation 3-1. The binding energy of the electron that was ejected is
comparatively small and can be ignored.
Eo 5 Esc 1 Ee
[3-1]
Compton scattering results in the ionization of the atom and a division of the
incident photons energy between the scattered photon and the ejected electron. The
ejected electron will lose its kinetic energy via excitation and ionization of atoms in
the surrounding material. The Compton scattered photon may traverse the medium
without interaction or may undergo subsequent interactions such as Compton
scattering, photoelectric absorption (to be discussed shortly), or Rayleigh scattering.
The energy of the scattered photon can be calculated from the energy of the incident photon and the angle (with respect to the incident trajectory) of the scattered
photon:
40
FIGURE 3-7 Compton scattering. The diagram shows the

incident photon with energy E0,
interacting with a valence-shell
electron that results in the ejection of the Compton electron
(Ee2) and the simultaneous emission of a Compton scattered
photon Esc emerging at an angle
relative to the trajectory of the
incident photon. K, L, and M are
electron shells.
Compton scattering
Valence electrons
Incident
photon
(E0)
Compton
electron (Ee-)
1 < 2
E sc
Angle of deflection
Scattered
photon (Esc)
Eo

Eo
1
(1 cos )
511 keV
[3-2]
where Esc 5 the energy of the scattered photon, Eo 5 the incident photon energy, and
5 the angle of the scattered photon.
As the incident photon energy increases, both scattered photons and electrons
are scattered more toward the forward direction (Fig. 3-8). In x-ray transmission
imaging, these photons are much more likely to be detected by the image receptor.
In addition, for a given scattering angle, the fraction of energy transferred to the
scattered photon decreases with increasing incident photon energy. Thus, for higher
energy incident photons, the majority of the energy is transferred to the scattered
electron. For example, for a 60-degree scattering angle, the scattered photon energy
(Esc) is 90% of the incident photon energy (Eo) at 100 keV but only 17% at 5MeV.
100
90
Relative probability (%)
FIGURE 3-8 Graph illustrates relative Compton scatter probability as a function of scattering angle for 20-, 80-, and 140-keV photons in
tissue. Each curve is normalized to 100%. (From
Bushberg JT. The AAPM/RSNA physics tutorial
for residents. X-ray interactions. RadioGraphics
1998;18:457468, with permission.)
80
70
60
140 keV
50
80 keV
40
20 keV
30
20
10
0
30
60
90
120
150
Scatter angle (degrees)
180
41
When Compton scattering occurs at the lower x-ray energies used in diagnostic
imaging (15 to 150 keV), the majority of the incident photon energy is transferred to
the scattered photon. For example, following the Compton interaction of an 80-keV
photon, the minimum energy of the scattered photon is 61 keV. Thus, even with
maximal energy loss, the scattered photons have relatively high energies and tissue
penetrability. In x-ray transmission imaging and nuclear emission imaging, the detection of scattered photons by the image receptors results in a degradation of image
contrast and an increase in random noise. These concepts, and many others related
to image quality, will be discussed in Chapter 4.
The laws of conservation of energy and momentum place limits on both scattering
angle and energy transfer. For example, the maximal energy transfer to the Compton
electron (and thus, the maximum reduction in incident photon energy) occurs with a
180-degree photon scatter (backscatter). In fact, the maximal energy of the scattered
photon is limited to 511 keV at 90 degrees scattering and to 255 keV for a 180-degree
scattering event. These limits on scattered photon energy hold even for extremely
high-energy photons (e.g., therapeutic energy range). The scattering angle of the
ejected electron cannot exceed 90 degrees, whereas that of the scattered photon can
be any value including a 180-degree backscatter. In contrast to the scattered photon,
the energy of the ejected electron is usually absorbed near the scattering site.
The incident photon energy must be substantially greater than the electrons
binding energy before a Compton interaction is likely to take place. Thus, the relative probability of a Compton interaction increases, compared to Rayleigh scattering
or photoelectric absorption, as the incident photon energy increases. The probability
of Compton interaction also depends on the electron density (number of electrons/g
3 density). With the exception of hydrogen, the total number of electrons/g is fairly
constant in tissue; thus, the probability of Compton scattering per unit mass is
nearly independent of Z, and the probability of Compton scattering per unit volume is approximately proportional to the density of the material. Compared to other
elements, the absence of neutrons in the hydrogen atom results in an approximate
doubling of electron density. Thus, hydrogenous materials have a higher probability
of Compton scattering than anhydrogenous material of equal mass.
The Photoelectric Effect

In the photoelectric effect, all of the incident photon energy is transferred to an electron, which is ejected from the atom. The kinetic energy of the ejected photoelectron
(Epe) is equal to the incident photon energy (Eo) minus the binding energy of the
orbital electron (Eb) (Fig. 3-9 left).

Epe 5 Eo 2 Eb
[3-3]
In order for photoelectric absorption to occur, the incident photon energy must
be greater than or equal to the binding energy of the electron that is ejected. The
ejected electron is most likely one whose binding energy is closest to, but less than,
the incident photon energy. For example, for photons whose energies exceed the
K-shell binding energy, photoelectric interactions with K-shell electrons are most
probable. Following a photoelectric interaction, the atom is ionized, with an innershell electron vacancy. This vacancy will be filled by an electron from a shell with
a lower binding energy. This creates another vacancy, which, in turn, is filled by an
electron from an even lower binding energy shell. Thus, an electron cascade from
outer to inner shells occurs. The difference in binding energy is released as either
characteristic x-rays or Auger electrons (see Chapter 2). The probability of characteristic x-ray emission decreases as the atomic number of the absorber decreases, and
42

Binding energy (keV)
100 keV
incident
photon
K
L
1 < 2 < 3 < 4
M
N
K
L
33
M
N
~0
33
67 keV photoelectron
~0
Characteristic
X-rays:
A: 1 keV (NM)
B: 4 keV (ML)
C: 28 keV (LK)
FIGURE 3-9 Photoelectric absorption. Left. The diagram shows that a 100-keV photon is undergoing photoelectric absorption with an iodine atom. In this case, the K-shell electron is ejected with a kinetic energy
equal to the difference (67 keV) between the incident photon energy (100 keV) and the K-shell binding energy
(33 keV). Right. The vacancy created in the K shell results in the transition of an electron from the L shell to the
K shell. The difference in their binding energies (i.e., 33 and 5 keV) results in a 28-keV Ka characteristic x-ray.
This electron cascade will continue, resulting in the production of other characteristic x-rays of lower energies.
Note that the sum of the characteristic x-ray energies equals the binding energy of the ejected photoelectron
(33 keV). Although not shown on this diagram, Auger electrons of various energies could be emitted in lieu
of the characteristic x-ray emissions.
thus, characteristic x-ray emission does not occur frequently for diagnostic energy
photon interactions in soft tissue. The photoelectric effect can and does occur with
valence shell electrons such as when light photons strike the high Z materials that
comprise the photocathode (e.g., cesium, rubidium and antimony) of a photomultiplier tube. These materials are specially selected to provide weakly bound electrons
(i.e., electrons with a low work function), so when illuminated the photocathode
readily releases electrons (see Chapter 17). In this case, no inner shell electron cascade occurs and thus no characteristic x-rays are produced.
Example: The K- and L-shell electron binding energies of iodine are 33 and 5 keV, respectively.
If a 100-keV photon is absorbed by a K-shell electron in a photoelectric interaction, the photoelectron is ejected with a kinetic energy equal to Eo 2 Eb 5 100 2 33 5 67 keV. A characteristic
x-ray or Auger electron is emitted as an outer-shell electron fills the K-shell vacancy (e.g., L to K
transition is 33 2 5 5 28 keV). The remaining energy is released by subsequent cascading events in
the outer shells of the atom (i.e., M to L and N to M transitions). Note that the total of all the characteristic x-ray emissions in this example equals the binding energy of the K-shell photoelectron
(Fig. 3-9, right).
Thus, photoelectric absorption results in the production of
1. A photoelectron
2. A positive ion (ionized atom)
3. Characteristic x-rays or Auger electrons
The probability of photoelectric absorption per unit mass is approximately proportional to Z3/E3, where Z is the atomic number and E is the energy of the incident
photon. For example, the photoelectric interaction probability in iodine (Z 5 53) is
(53/20)3 or 18.6 times greater than in calcium (Z 5 20) for a photon of a particular
energy.
Mass attenuation coefficient (cm2/g)
100
Photoelectric effect
30
10
3
K-edge
1
Iodine
0.3
0.01
43
FIGURE 3-10 Photoelectric mass attenuation

coefficients for tissue (Zeffective 5 7), and iodine
(Z 5 53) as a function of energy. Abrupt increase
in the attenuation coefficients called absorption edges occur due to increased probability of
photoelectric absorption when the photon energy
just exceeds the binding energy of inner-shell
electrons (e.g., K, L, M,), thus increasing the
number of electrons available for interaction. This
process is very significant in high-Z elements, such
as iodine and barium, for x-rays in the diagnostic
energy range.
Tissue
0.03
0.01
20
40
60
80 100 120
X-ray energy (keV)
140
The benefit of photoelectric absorption in x-ray transmission imaging is that

there are no scattered photons to degrade the image. The fact that the probability of
photoelectric interaction is proportional to 1/E3 explains, in part, why image contrast
decreases when higher x-ray energies are used in the imaging process (see Chapters
4 and 7). If the photon energies are doubled, the probability of photoelectric interaction is decreased eightfold: ()3 5 1/8.
Although the probability of the photoelectric effect decreases, in general, with
increasing photon energy, there is an exception. For every element, the probability of
the photoelectric effect, as a function of photon energy, exhibits sharp discontinuities
called absorption edges (see Fig. 3-10). The probability of interaction for photons of
energy just above an absorption edge is much greater than that of photons of energy
slightly below the edge. For example, a 33.2-keV x-ray photon is about six times as
likely to have a photoelectric interaction with an iodine atom as a 33.1-keV photon.
As mentioned above, a photon cannot undergo a photoelectric interaction with an
electron in a particular atomic shell or subshell if the photons energy is less than the
binding energy of that shell or subshell. This causes the dramatic decrease in the probability of photoelectric absorption for photons whose energies are just below the binding energy of a shell. Thus, the photon energy corresponding to an absorption edge
is the binding energy of the electrons in that particular shell or subshell. An absorption edge is designated by a letter, representing the atomic shell of the electrons, followed by a roman numeral subscript denoting the subshell (e.g., K, LI, LII, LIII).
The photon energy corresponding to a particular absorption edge increases with the
atomic number (Z) of the element. For example, the primary elements comprising soft
tissue (H, C, N, and O) have absorption edges below 1 keV. The element iodine (Z 5 53)
commonly used in radiographic contrast agents to provide enhanced x-ray attenuation,
has a K-absorption edge of 33.2 keV (Fig. 3-10). The K-edge energy of the target material
in most x-ray tubes (tungsten, Z 5 74) is 69.5 keV. The K- and L-shell binding energies
for elements with atomic numbers 1 to 100 are provided in Appendix C, Table C-3.
The photoelectric process predominates when lower energy photons interact
with high Z materials (Fig. 3-11). In fact, photoelectric absorption is the primary
mode of interaction of diagnostic x-rays with image receptors, radiographic contrast
materials, and radiation shielding, all of which have much higher atomic numbers
than soft tissue. Conversely, Compton scattering predominates at most diagnostic
and therapeutic photon energies in materials of lower atomic number such as tissue
and air. At photon energies below 50 keV, photoelectric interactions in soft tissue

Diagnostic
Radiology
20
Nuclear
Medicine
(70-80)
511
100
100
keV
Percent Compton scatter
0
10
50
ue
Tiss
50
Lea
Na l
Bon
FIGURE 3-11 Graph of the percentage of contribution of photoelectric (left scale) and Compton (right
scale) attenuation processes for various materials as a
function of energy. When diagnostic energy photons
(i.e., diagnostic x-ray effective energy of 20 to 80 keV;
nuclear medicine imaging photons of 70 to 511 keV)
interact with materials of low atomic number (e.g., soft
tissue), the Compton process dominates.
Percent photoelectric absorption
44
100
1000
play an important role in medical imaging. The photoelectric absorption process can
be used to amplify differences in attenuation between tissues with slightly different
atomic numbers, thereby improving image contrast. This differential absorption is
exploited to improve image contrast through the selection of x-ray tube target material and filters in mammography (see Chapter 8).
Pair Production
Pair production can only occur when the energies of x-rays and gamma rays exceed
1.02 MeV. In pair production, an x-ray or gamma ray interacts with the electric field of
the nucleus of an atom. The photons energy is transformed into an electron-positron
pair (Fig. 3-12A). The rest mass energy equivalent of each electron is 0.511 MeV,
and this is why the energy threshold for this reaction is 1.02 MeV. Photon energy in
excess of this threshold is imparted to the electron (also referred to as a negatron or
beta minus particle) and positron as kinetic energy. The electron and positron lose
their kinetic energy via excitation and ionization. As discussed previously, when the
positron comes to rest, it interacts with a negatively charged electron, resulting in the
formation of two oppositely directed 0.511-MeV annihilation photons (Fig. 3-12B).
Pair production does not occur in diagnostic x-ray imaging because the threshold
photon energy is well beyond even the highest energies used in medical imaging. In
fact, pair production does not become significant until the photon energies greatly
exceed the 1.02-MeV energy threshold.
3.3 Attenuation of x-rays and Gamma Rays

Attenuation is the removal of photons from a beam of x-rays or gamma rays as it
passes through matter. Attenuation is caused by both absorption and scattering of
the primary photons. The interaction mechanisms discussed in the previous section,
in varying degrees, cause the attenuation. At low photon energies (less than 26 keV),
the photoelectric effect dominates the attenuation processes in soft tissue. However,
as previously discussed, the probability of photoelectric absorption is highly dependent on photon energy and the atomic number of the absorber. When higher energy
photons interact with low Z materials (e.g., soft tissue), Compton scattering dominates (Fig.3-13). Rayleigh scattering occurs in medical imaging with low probability,
45
Incident
photon
Excitation and
ionization
(Negatron)
+
(Positron)
Excitation
and
ionization
0.511 MeV
0.511 MeV
Annihilation Radiation
~180
FIGURE 3-12 Pair production. A. The diagram illustrates the pair production process in which a high-energy
incident photon, under the influence of the atomic nucleus, is converted to an electron-positron pair. Both
electrons (positron and negatron) expend their kinetic energy by excitation and ionization in the matter they
traverse. B. However, when the positron comes to rest, it combines with an electron producing the two 511keV annihilation radiation photons. K, L, and M are electron shells.
c omprising about 10% of the interactions in mammography and 5% in chest radiography. Only at very high photon energies (greater than 1.02 MeV), well beyond the range
of diagnostic and nuclear radiology, does pair production contribute to attenuation.
Linear Attenuation Coefficient

The fraction of photons removed from a monoenergetic beam of x-rays or gamma rays
per unit thickness of material is called the linear attenuation coefficient (m), typically
expressed in units of inverse centimeters (cm21). The number of photons removed
from the beam traversing a very small thickness Dx can be expressed as

n 5 m N Dx
[3-4]
where n 5 the number of photons removed from the beam, and N 5 the number of
photons incident on the material.
For example, for 100-keV photons traversing soft tissue, the linear attenuation
coefficient is 0.016 mm21. This signifies that for every 1,000 monoenergetic photons
incident upon a 1-mm thickness of tissue, approximately 16 will be removed from
the beam, by either absorption or scattering.
As the thickness increases, however, the relationship is not linear. For example,
it would not be correct to conclude from Equation 3-4 that 6 cm of tissue would
attenuate 960 (96%) of the incident photons. To accurately calculate the number of
FIGURE 3-13 Graph of the Rayleigh,

photoelectric, Compton, pair production,
and total mass attenuation coefficients for
soft tissue (Z 7) as a function of photon
energy.
Mass Attenuation Coefficients for Soft Tissue

10
Mass attenuation coefficient (cm2/g)
46
3
1
Total
0.3
0.1
Photoelectric
0.03
0.01
Compton
Rayleigh
Pair production
0.003
0.001
10
100
1,000
Energy (keV)
10,000
photons removed from the beam using Equation 3-4, multiple calculations utilizing
very small thicknesses of material (Dx) would be required. Alternatively, calculus
can be employed to simplify this otherwise tedious process. For a monoenergetic
beam of photons incident upon either thick or thin slabs of material, an exponential
relationship exists between the number of incident photons (N0) and those that are
transmitted (N) through a thickness x without interaction:

N 5 N0e2x
[3-5]
Thus, using the example above, the fraction of 100-keV photons transmitted
through 6 cm of tissue is
1
N /N 0 e(0.16 cm
)(6 cm)
0.38
This result indicates that, on average, 380 of the 1,000 incident photons (i.e.,
38%) would be transmitted through the 6-cm slab of tissue without interacting.
Thus, the actual attenuation (1 0.38 or 62%) is much lower than would have been
predicted from Equation 3-4.
The linear attenuation coefficient is the sum of the individual linear attenuation
coefficients for each type of interaction:

5 Rayleigh 1 photoelectric effect 1 Compton scatter 1 pair production
[3-6]
In the diagnostic energy range, the linear attenuation coefficient decreases with
increasing energy except at absorption edges (e.g., K-edge). The linear attenuation
coefficient for soft tissue ranges from approximately 0.35 to 0.16 cm21 for photon
energies ranging from 30 to 100 keV.
For a given thickness of material, the probability of interaction depends on the
number of atoms the x-rays or gamma rays encounter per unit distance. The density (r, in g/cm3) of the material affects this number. For example, if the density is
doubled, the photons will encounter twice as many atoms per unit distance through
the material. Thus, the linear attenuation coefficient is proportional to the density of
the material, for instance:
water ice water vapor
The relationship among material density, electron density, electrons per mass, and the
linear attenuation coefficient (at 50 keV) for several materials is shown in Table3-1.
47
TABLE 3-1MATERIAL DENSITY, ELECTRONS PER MASS, ELECTRON
ENSITY, AND THE LINEAR ATTENUATION COEFFICIENT

D
(AT 50 keV) FOR SEVERAL M
ATERIALS
MATERIAL
DENSITY (g/cm3)
ELECTRONS PER
MASS (e/g) 3 1023
ELECTRON DENSITY
(e/cm3) 3 1023
@ 50 keV (cm21)
Hydrogen gas
0.000084
5.97
0.0005
0.000028
Water vapor
0.000598
3.34
0.002
0.000128
Air
0.00129
3.006
0.0038
0.000290
Fat
0.91
3.34
3.04
0.193
Ice
0.917
3.34
3.06
0.196
Water
3.34
3.34
0.214
Compact bone
1.85
3.192
5.91
0.573
Mass Attenuation Coefficient

For a given material and thickness, the probability of interaction is proportional to
the number of atoms per volume. This dependency can be overcome by normalizing
the linear attenuation coefficient for the density of the material. The linear attenuation coefficient, normalized to unit density, is called the mass attenuation coefficient.
Mass Attenuation Coefficient ( / ) cm2 / g
Linear Attenuation Coefficient (m )cm1 ]

Density of Material (r ) [g/cm 3 ]
[3-7]
The linear attenuation coefficient is usually expressed in units of cm21, whereas

the units of the mass attenuation coefficient are usually cm2/g.
The mass attenuation coefficient is independent of density. Therefore, for a given
photon energy,
water/water 5 ice/ice 5 water vapor/water vapor
However, in radiology, we do not usually compare equal masses. Instead, we usually compare regions of an image that correspond to irradiation of adjacent volumes
of tissue. Therefore, density, the mass contained within a given volume, plays an
important role. Thus, one can radiographically visualize ice in a cup of water due to
the density difference between the ice and the surrounding water (Fig. 3-14).
To calculate the linear attenuation coefficient for a density other than 1 g/cm3, the
density of the material is multiplied by the mass attenuation coefficient to yield the linear
attenuation coefficient. For example, the mass attenuation coefficient of air, for 60-keV
photons, is 0.186 cm2/g. At typical room conditions, the density of air is 0.00129g/cm3.
Therefore, the linear attenuation coefficient of air under these c onditions is
5 (/o) 5 (0.186 cm2/g) (0.00129 g/cm3) 5 0.000240 cm21
To use the mass attenuation coefficient to compute attenuation, Equation 3-5 can
be rewritten as

N N oe
[3-8]
Because the use of the mass attenuation coefficient is so common, scientists in

this field tend to think of thickness not as a linear distance x (in cm) but rather in
48
FIGURE 3-14 Radiograph (acquired

at 125 kV with an antiscatter grid) of
two ice cubes in a plastic container of
water. The ice cubes can be visualized because of their lower electron
density relative to that of liquid water.
The small radiolucent objects seen at
several locations are the result of air
bubbles in the water. (From Bushberg
JT. The AAPM/RSNA physics tutorial for residents. X-ray interactions.
RadioGraphics 1998;18:457468, with
permission.)
terms of mass per unit area rx (in g/cm2). The product rx is called the mass thickness
or areal thickness.
Half-Value Layer
The half-value layer (HVL) is defined as the thickness of material required to reduce
the intensity (e.g., air kerma rate) of an x-ray or gamma-ray beam to one half of
its initial value. The HVL of a beam is an indirect measure of the photon energies
(also referred to as the quality) of a beam, when measured under conditions of narrow-beam geometry. Narrow-beam geometry refers to an experimental configuration
that is designed to exclude scattered photons from being measured by the detector
(Fig. 3-15A). In broad-beam geometry, the beam is sufficiently wide that a substantial
fraction of scattered photons remain in the beam. These scattered photons reaching
the detector (Fig. 3-15B) result in an underestimation of the attenuation coefficient
Collimator
Collimator
Attenuator
Attenuator
Detector
Source
X
Detector
Source
X
Some Photons are

Scattered into the
Detector
Scattered
Photons not
Detected
Narrow-Beam Geometry
Broad-Beam Geometry
FIGURE 3-15 A. Narrow-beam geometry means that the relationship between the source shield and the
detector is such that almost no scattered photons interact with the detector. B. In broad-beam geometry, scattered
photons may reach the detector; thus, the measured attenuation is less compared with narrow-beam conditions.
49
(i.e., an overestimated HVL). Most practical applications of attenuation (e.g., patient

imaging) occur under broad-beam conditions. The tenth-value layer (TVL) is analogous to the HVL, except that it is the thickness of material that is necessary to reduce
the intensity of the beam to a tenth of its initial value. The TVL is often used in x-ray
room shielding design calculations (see Chapter 21).
For monoenergetic photons under narrow-beam geometry conditions, the probability of attenuation remains the same for each additional HVL thickness placed in
the beam. Reduction in beam intensity can be expressed as ()n, where n equals the
number of HVLs. For example, the fraction of monoenergetic photons transmitted
through 5 HVLs of material is
3 3 3 3 5 ()5 5 1/32 5 0.031 or 3.1%
Therefore, 97% of the photons are attenuated (removed from the beam). The
HVL of a diagnostic x-ray beam, measured in millimeters of aluminum under narrowbeam conditions, is a surrogate measure of the penetrability of an x-ray spectrum.
It is important to understand the relationship between m and HVL. In
Equation 3-5, N is equal to No/2 when the thickness of the absorber is 1 HVL. Thus,
for a monoenergetic beam,
No/2 5 Noe2m(HVL)
1/2 5 e2m(HVL)
ln (1/2) 5 ln e2m(HVL)
20.693 5 2m (HVL)
HVL 5 0.693/
[3-9]
For a monoenergetic incident photon beam, the HVL can be easily calculated
from the linear attenuation coefficient, and vice versa. For example, given
1. m 5 0.35 cm21
HVL 5 0.693/0.35 cm21 5 1.98 cm
2. HVL 5 2.5 mm 5 0.25 cm
m 5 0.693/0.25 cm 5 2.8 cm21
The HVL and m can also be calculated if the percent transmission is measured under
narrow-beam geometry.
Example: If a 0.2-cm thickness of material transmits 25% of a monoenergetic beam of photons, calculate the HVL of the beam for that material.
Step1.
0.25 5 e2m(0.2 cm)
Step 2. ln 0.25 5 2m(0.2 cm)

Step 3. m 5 (2ln 0.25)/(0.2 cm) 5 6.93 cm21
Step 4. HVL 5 0.693/m 5 0.693/6.93 cm21 5 0.1 cm
HVLs for photons from three commonly used diagnostic radionuclides (201Tl, 99mTc,
and 18F) are listed for tissue and lead in Table 3-2. Thus, the HVL is a function of (a)
photon energy, (b) geometry, and (c) attenuating material.
Effective Energy
X-ray beams in radiology are polyenergetic, meaning that they are composed of a
spectrum of x-ray energies. The determination of the HVL in diagnostic radiology is a
way of characterizing the penetrability of the x-ray beam. The HVL, usually measured
50
TABLE 3-2HVLs OF TISSUE, ALUMINUM, AND LEAD FOR
X-RAYS AND GAMMA RAYS COMMONLY

USED IN NUCLEAR MEDICINE
HALF VALUE LAYER (mm)
PHOTON SOURCE
TISSUE
LEAD
70 keV x-rays (201Tl)
37
0.2
140 keV g-rays (99mTc)
44
0.3
511 keV g-rays ( F)
75
4.1
18
Tc, technetium; Tl, thallium; Fl, fluorine.

Note: These values are based on the narrow-beam geometry attenuation and
neglecting the effect of scatter. Shielding calculations (discussed in Chapter 21)
are typically for broad-beam conditions.
in millimeters of aluminum (mm Al) in diagnostic radiology, can be converted to a

quantity called the effective energy. The effective energy of a polyenergetic x-ray beam
is an estimate of the penetration power of the x-ray beam, expressed as the energy
of a monoenergetic beam that would exhibit the same effective penetrability. The
relationship between HVL (in mm Al) and effective energy is given in Table 3-3. The
effective energy of an x-ray beam from a typical diagnostic x-ray tube is one third to
one half the maximal value.
Mean Free Path

One cannot predict the range of a single photon in matter. In fact, the range can vary
from zero to infinity. However, the average distance traveled before interaction can
TABLE 3-3HVL AS A FUNCTION OF THE
FFECTIVE ENERGY OF AN
E
X-RAY BEAM
HVL (mm Al)
EFFECTIVE ENERGY (keV)
0.26
14
0.75
20
1.25
24
1.90
28
3.34
35
4.52
40
5.76
45
6.97
50
9.24
60
11.15
70
12.73
80
14.01
90
15.06
100
Al, aluminum.
51
Average Photon Energy and HVL Increases

Photon Intensity (i.e. quantity) decreases
FIGURE 3-16 Beam hardening results from preferential absorption of lower energy photons as the x-rays
traverse matter.
be calculated from the linear attenuation coefficient or the HVL of the beam. This
length, called the mean free path (MFP) of the photon beam, is
MFP
1
1
1.44 HVL
0.693 / HVL
[3-10]
Beam Hardening
The lower energy photons of the polyenergetic x-ray beam will preferentially be
removed from the beam while passing through matter. The shift of the x-ray spectrum to higher effective energies as the beam transverses matter is called beam hardening (Fig. 3-16). Low-energy (soft) x-rays will not penetrate the entire thickness of
the body; thus, their removal reduces patient dose without affecting the diagnostic
quality of the exam. X-ray machines remove most of this soft radiation with filters, thin plates of aluminum, copper, or other materials placed in the beam. This
added filtration will result in an x-ray beam with a higher effective energy and thus
a greater HVL.
The homogeneity coefficient is the ratio of the first to the second HVL and
describes the polyenergetic character of the beam. The first HVL is the thickness
that reduces the incident intensity to 50%, and the second HVL reduces it to 25%
of its original intensity (i.e., 0.5 3 0.5 5 0.25). For most of diagnostic x-ray imaging the homogeneity coefficient of the x-ray spectrum is between 0.50.7. However
for special applications such as conventional projection mammography with factors
optimized to enhance the spectral uniformity of the x-ray beam the homogeneity
coefficient can be as high as 0.97. A monoenergetic source of gamma rays has a
homogeneity coefficient equal to 1.
The maximal x-ray energy of a polyenergetic spectrum can be estimated by monitoring the homogeneity coefficient of two heavily filtered beams (e.g., 15th and 16th
HVLs). As the coefficient approaches 1, the beam is essentially monoenergetic. Measuring m for the material in question under heavy filtration conditions and matching
it to known values of m for monoenergetic beams provides an approximation of the
maximal energy.
52
3.4 Absorption of Energy from X-rays and Gamma Rays
Radiation Units and Measurements

The International System of units (SI) provides a common system of units for science
and technology. The system consists of seven base units: meter (m) for length, kilogram (kg) for mass, second (s) for time, ampere (A) for electric current, kelvin (K)
for temperature, candela (cd) for luminous intensity, and mole (mol) for the amount
of substance. In addition to the seven base units, there are derived units defined as
combinations of the base units. Examples of derived units are speed (m/s) and density (kg/m3). Details regarding derived units used in the measurement and calculation of radiation dose for specific applications can be found in the documents of the
International Commission on Radiation Units and Measurements (ICRU) and the
International Commission on Radiological Protection (ICRP). Several of these units
are described below, whereas others related to specific imaging modalities or those
having specific regulatory significance are described in the relevant chapters.
Fluence, Flux, and Energy Fluence

The number of photons or particles passing through a unit cross-sectional area is
referred to as the fluence and is typically expressed in units of cm22. The fluence is
given the symbol F.
Photons
=

[3-11]
Area
The fluence rate (e.g., the rate at which photons or particles pass through a unit
, is simply the
area per unit time) is called the flux. The flux, given the symbol
fluence per unit time.

Photons

Area Time
[3-12]
The flux is useful in situations in which the photon beam is on for extended periods of time, such as in fluoroscopy. Flux has the units of cm22 s21.
The amount of energy passing through a unit cross-sectional area is referred to as
the energy fluence. For a monoenergetic beam of photons, the energy fluence () is
simply the product of the fluence (F) and the energy per photon (E).

Photons
Energy

Area
Photon
[3-13]
The units of are energy per unit area, J m22, or joules per m2. For a polyenergetic spectrum, the total energy in the beam is tabulated by multiplying the number
of photons at each energy by that energy and adding these products. The energy fluence rate or energy flux is the energy fluence per unit time.
Kerma
As a beam of indirectly (uncharged) ionizing radiation (e.g., x-rays or gamma rays or neutrons) passes through a medium, it deposits energy in the medium in a two-step process:
Step 1. Energy carried by the photons (or other indirectly ionizing radiation) is transformed into kinetic energy of charged particles (such as electrons). In the case of
x-rays and gamma rays, the energy is transferred by photoelectric absorption, Compton scattering, and, for very high energy photons, pair production.
53
Step 2. The directly ionizing (charged) particles deposit their energy in the medium by
excitation and ionization. In some cases, the range of the charged particles is sufficiently
large that energy deposition is some distance away from the initial interactions.
Kerma (K) is an acronym for kinetic energy released in matter. Kerma is defined
at the kinetic energy transferred to charged particles by indirectly ionizing radiation per unit mass, as described in Step 1 above. The SI unit of Kerma is the joule
per kilogram with the special name of the gray (Gy) or milligray (mGy), where
1 Gy 5 1 J kg21. For x-rays and gamma rays, kerma can be calculated from the mass
energy transfer coefficient of the material and the energy fluence.
Mass Energy Transfer Coefficient

The mass energy transfer coefficient is given the symbol:
tr
o
The mass energy transfer coefficient is the mass attenuation coefficient multiplied
by the fraction of the energy of the interacting photons that is transferred to charged
particles as kinetic energy. As was mentioned above, energy deposition in matter by
photons is largely delivered by the energetic charged particles produced by photon
interactions. The energy in scattered photons that escape the interaction site is not
transferred to charged particles in the volume of interest. Furthermore, when pair
production occurs, 1.02 MeV of the incident photons energy is required to produce
the electron-positron pair and only the remaining energy (Ephoton 2 1.02 MeV) is
given to the electron and positron as kinetic energy. Therefore, the mass energy transfer coefficient will always be less than the mass attenuation coefficient.
For 20-keV photons in tissue, for example, the ratio of the energy transfer coefficient
to the attenuation coefficient (mtr/m) is 0.68, but this reduces to 0.18 for 50-keV photons,
as the amount of Compton scattering increases relative to photoelectric absorption.
Calculation of Kerma
For a monoenergetic photon beam with an energy fluence and energy E, the kerma
K is given by

K = otr
[3-14]

E

where tr is the mass energy transfer coefficient of the absorber at energy E. The SI
o E
units of energy fluence are J m22, and the SI units of the mass energy transfer coefficient
are m2 kg21, and thus their product, kerma, has units of J kg21 (1 J kg21 5 1 Gy).
Absorbed Dose
The quantity absorbed dose (D) is defined as the energy (E) imparted by ionizing
radiation per unit mass of irradiated material (m):
E

[3-15]
m
Unlike Kerma, absorbed dose is defined for all types of ionizing radiation (i.e.,
directly and indirectly ionizing). However, the SI unit of absorbed dose and kerma,
is the same (gray), where 1 Gy 5 1 J kg21. The older unit of absorbed dose is the

D=
54
rad (an acronym for radiation absorbed dose). One rad is equal to 0.01 J kg21. Thus,
there are 100 rads in a gray, and 1 rad 5 10 mGy.
If the energy imparted to charged particles is deposited locally and the bremsstrahlung produced by the energetic electrons is negligible, the absorbed dose will be equal
to the kerma. For x-rays and gamma rays, the absorbed dose can be calculated from
the mass energy absorption coefficient and the energy fluence of the beam.
Mass Energy Absorption Coefficient

The mass energy transfer coefficient discussed above describes the fraction of the mass
attenuation coefficient that gives rise to the initial kinetic energy of electrons in a small
volume of absorber. The mass energy absorption coefficient will be the same as the
mass energy transfer coefficient when all transferred energy is locally absorbed. However, energetic electrons may subsequently produce bremsstrahlung radiation (x-rays),
which can escape the small volume of interest. Thus, the mass energy absorption coefficient may be slightly smaller than the mass energy transfer coefficient. For the energies
used in diagnostic radiology and for low-Z absorbers (air, water, tissue), the amount of
radiative losses (bremsstrahlung) is very small. Thus, for diagnostic radiology,
en tr
o o
The mass energy absorption coefficient is useful when energy deposition calculations are to be made.
Calculation of Dose
The distinction between kerma and dose is slight for the relatively low x-ray energies
used in diagnostic radiology. The dose in any material is given by

D = oen
[3-16]
The difference between the calculation of kerma and dose for air is that kerma is
defined using the mass energy transfer coefficient, whereas dose is defined using the
mass energy absorption coefficient. The mass energy transfer coefficient defines the
energy transferred to charged particles, but these energetic charged particles (mostly
electrons) in the absorber may experience radiative losses, which can exit the small
volume of interest. The coefficient
en
o
E
takes into account the radiative losses, and thus
tr
en
o o
E
E
Exposure
The amount of electrical charge (Q) produced by ionizing electromagnetic radiation
per mass (m) of air is called exposure (X):

X=
Q

m
[3-17]
55
Exposure is expressed in the units of charge per mass, that is, coulombs per kg
(C kg21). The historical unit of exposure is the roentgen (abbreviated R), which is
defined as
1 R 5 2.58 3 1024 C kg21 (exactly)
Radiation beams are often expressed as an exposure rate (R/h or mR/min). The
output intensity of an x-ray machine can be measured and expressed as an exposure
(R) per unit of current times exposure duration (milliampere second or mAs) under
specified operating conditions (e.g., 5 mR/mAs at 70 kV for a source-image distance
of 100 cm, and with an x-ray beam filtration equivalent to 2 mm Al).
Exposure is a useful quantity because ionization can be directly measured with
air-filled radiation detectors, and the effective atomic numbers of air and soft tissue
are approximately the same. Thus, exposure is nearly proportional to dose in soft
tissue over the range of photon energies commonly used in radiology. However, the
quantity of exposure is limited in that it applies only to the interaction of ionizing
photons (not charged particle radiation) in air (not any other substance).
The exposure can be calculated from the dose to air. Let the ratio of the dose (D)
to the exposure (X) in air be W and substituting in the above expressions for D and
X yields
D E
W= =

[3-18]

X Q
W, the average energy deposited per ion pair in air, is approximately constant as a
function of energy. The value of W is 33.85 eV/ion pair or 33.85 J/C.
In terms of the traditional unit of exposure, the roentgen, the dose to air is
2.58 104 C
K air = W X
kg R
or

Gy
K air (Gy) = 0.00873 X
R
[3-19]
Thus, one R of exposure results in 8.73 mGy of air dose. The quantity exposure is
still in common use in the United States, but the equivalent SI quantities of air dose
or air kerma are used exclusively in most other countries. These conversions can be
simplified to

K air (mGy) =
X (mR)

114.5(mR / mGy )
[3-20]
K air ( Gy)
X (R)
114.5(R / Gy )
[3-21]
3.5 Imparted Energy, Equivalent Dose, and Effective Dose
Imparted Energy
The total amount of energy deposited in matter, called the imparted energy (), is the
product of the dose and the mass over which the energy is imparted. The unit of
imparted energy is the joule.

5 (J/kg) 3 kg 5 J
[3-22]
56
For example, assume a head computed tomography (CT) scan delivers a 30 mGy
dose to the tissue in each 5-mm slice. If the scan covers 15 cm, the dose to the irradiated volume is 30 mGy; however, the imparted (absorbed) energy is approximately
15 times that in a single scan slice.
Other modality-specific dosimetric quantities such as the CT dose index (CTDI),
the dose-length product (DLP), multiple scan average dose (MSAD) used in computed tomography, and the entrance skin dose (ESD) and kerma-area-product (KAP)
used in fluoroscopy will be introduced in context with their imaging modalities and
discussed again in greater detail in Chapter 11. Mammography-specific dosimetric
quantities such as the average glandular dose (AGD) are presented in Chapter 8.
The methods for calculating doses from radiopharmaceuticals used in nuclear
imaging utilizing the Medical Internal Radionuclide Dosimetry (MIRD) scheme are
discussed in Chapter 16. Dose quantities that are defined by regulatory agencies,
such as the total effective dose equivalent (TEDE) used by the Nuclear Regulatory
Commission (NRC), are presented in Chapter 21.
Equivalent Dose
Not all types of ionizing radiation cause the same biological damage per unit absorbed
dose. To modify the dose to reflect the relative effectiveness of the type of radiation
in producing biologic damage, a radiation weighting factor (wR) was established by the
ICRP as part of an overall system for radiation protection (see Chapter 21), High LET
radiations that produce dense ionization tracks cause more biologic damage per unit
dose than low LET radiations. This type of biological damage (discussed in greater
detail in chapter 20) can increase the probability of stochastic effects like cancer and
thus are assigned higher radiation weighting factors. The product of the absorbed
dose (D) and the radiation weighing factor is the equivalent dose (H).

H 5 D wR
[3-23]
The SI unit for equivalent dose is joule per kilogram with the special name of the
sievert (Sv), where 1 Sv 5 1 J kg21. Radiations used in diagnostic imaging (x-rays and
gamma rays) as well as the energetic electrons set into motion when these photons
interact with tissue or when electrons are emitted during radioactive decay, have a wR
TABLE 3-4RADIATION WEIGHTING FACTORS (wR) FOR
VARIOUS TYPES OF RADIATION

TYPE OF RADIATION
RADIATION WEIGHTING FACTOR (wR)
X-rays, gamma rays, beta

particles, and electrons
Protons
Neutrons (energy dependent)a
2.520
Alpha particles and other

multiple-charged particles
20
Note: For radiations principally used in medical imaging (x-rays and gamma
rays) and beta particles, wR 5 1; thus, the absorbed dose and equivalent dose
are equal (i.e., 1 Gy 5 1 Sv).
a
wR values are a continuous function of energy with a maximum of 20 at
approximately 1 Mev, minimum of 2.5 at 1 keV and 5 at 1 BeV). Adapted from
ICRP Publication 103, The 2007 Recommendations of the International Commission on Radiological Protection. Ann. ICRP 37 (24), Elsevier, 2008.
57
of 1: thus, 1 mGy 3 1 (wR) 5 1 mSv. For heavy charged particles such as alpha particles, the LET is much higher, and thus, the biologic damage and the associated wR
are much greater (Table 3-4). For example, 10 mGy from alpha radiation may have
the same biologic effectiveness as 200 mGy of x-rays.
The quantity H replaces an earlier but similar quantity, the dose equivalent, which
is the product of the absorbed dose and the quality factor (Q) (Equation 3-24). The
quality factor is similar to wR.

H 5 DQ
[3-24]
The need to present these out-of-date dose quantities arise because regulatory
agencies in the United States have not kept pace with current recommendations of
national and international organizations for radiation protection and measurement.
The traditional unit for both the dose equivalent and the equivalent dose is the rem.
A sievert is equal to 100 rem, and 1 rem is equal to 10 mSv.
Effective Dose
Biological tissues vary in sensitivity to the effects of ionizing radiation. Tissue weighting factors (wT) were also established by the ICRP as part of their radiation protection
TABLE 3-5TISSUE WEIGHTING FACTORS ASSIGNED
BY THE INTERNATIONAL COMMISSION ON

RADIOLOGICAL PROTECTION (ICRP Report 103)
ORGAN/TISSUE
wT
Breast
0.12
Bone marrow
0.12
Colona
0.12
Lung
0.12
Stomach
0.12
Remainder
0.12
Gonadsc
0.08
Bladder
0.04
Esophagus
0.04
Liver
0.04
Thyroid
0.04
Bone surface
0.01
Brain
0.01
Salivary gland
0.01
Skin
0.01
Total
1.0
% OF TOTAL DETRIMENT
72
16
100
The dose to the colon is taken to be the mass-weighted mean of upper and lower
large intestine doses.
b
Shared by remainder tissues (14 in total, 13 in each sex) are adrenals, extrathoracic
tissue, gallbladder, heart, kidneys, lymphatic nodes, muscle, oral mucosa, pancreas,
prostate (male), small intestine, spleen, thymus, uterus/cervix (female).
c
The wT for gonads is applied to the mean of the doses to testes and ovaries.
Adapted from ICRP Publication 103, The 2007 Recommendations of the International
Commission on Radiological Protection. Ann. ICRP 37 (24), Elsevier, 2008.
a
rem
A measure of equivalent dose, weighted for the bio- Sievert (Sv)

logical sensitivity of the exposed tissues and organs
(relative to whole body exposure) to stochastic health
effects in humans
Includes backscatter (discussed in Chapters 9 and 11).

ICRP, International Commission on Radiological Protection.
Effective dose (defined

by ICRP in 1990 to replace
effective dose equivalent
and modified in 2007
with different wT values)
rem
Sievert (Sv)
Effective dose e
quivalent A measure of dose equivalent, weighted for the
(defined by ICRP in 1977) biological sensitivity of the exposed tissues and organs
(relative to whole body exposure) to stochastic health
effects in humans
Joule (J)
rem
Total radiation energy imparted to matter
Imparted energy
Gray (Gy)
1 Gy 5 J kg21
Sievert (Sv)
Kinetic energy transferred to charged particles per

unit mass of air
Air kerma
Gray (Gy)
1 Gy 5 J kg21
rad
1 rad 5 0.01 J kg21
Equivalent dose (defined A measure of absorbed dose weighted for the biological effectiveness of the type(s) of radiation (relative to
by ICRP in 1990 to
replace dose equivalent) low LET photons and electrons) to produce stochastic
health effects in humans
Kinetic energy transferred to charged particles

per unit mass
Kerma
Gray (Gy)
1 Gy 5 J kg21
Roentgen (R)
rem
Amount of energy imparted by radiation per mass
Absorbed dose
C kg21
Sievert (Sv)
Amount of ionization per mass of

air due to x-rays and gamma rays
Exposure
Dose equivalent (defined A measure of absorbed dose weighted for the biologiby ICRP in 1977)
cal effectiveness of the type(s) of radiation (relative to
low LET photons and electrons) to produce stochastic
health effects in humans
DESCRIPTION OF QUANTITY
QUANTITY
SI UNITS
TRADITIONAL UNITS
(ABBREVIATIONS) (ABBREVIATIONS)
AND DEFINITIONS AND DEFINITIONS
HE
Kair
E 5 wT HT
HE 5 wTHT
H 5 wR D
1 rem 5 10 mSv
100 rem 5 1 Sv
H5QD
1 rem 5 10 mSv
100 rem 5 1 Sv
Dose (J kg21) 3 mass (kg) 5 J
1 mGy 5 0.115 R @ 30 kV
1 mGy 5 0.114 R @ 60 kV
1 mGy 5 0.113 R @ 100 kV
1 mGy 0.140 rad (dose to skin)a
1 mGy 1.4 mGy (dose to skin)a
1 rad 5 10 mGy
100 rad 51 Gy
1R 5 2.58 3 1024 C kg21

1R 5 8.708 mGy air kerma @ 30 kV
DEFINITIONS AND CONVERSION

SYMBOL FACTORS
TABLE 3-6 RADIOLOGICAL QUANTITIES, SYSTEM INTERNATIONAL (SI) UNITS, AND TRADITIONAL UNITS
59
system to assign a particular organ or tissue (T) the proportion of the detriment1 from
stochastic effects (e.g., cancer and hereditary effects, discussed further in Chapter 20)
resulting from irradiation of that tissue compared to uniform whole-body irradiation
(ICRP Publication 103, 2007). These tissue weighting factors are shown in Table 3-5.
The sum of the products of the equivalent dose to each organ or tissue irradiated
(HT) and the corresponding weighting factor (wT) for that organ or tissue is called the
effective dose (E).

E(Sv) 5 [wT 3 HT(Sv)]

T
[3-25]
The effective dose is expressed in the same units as the equivalent dose (sievert
orrem).
The wT values were developed for a reference population of equal numbers of both
genders and a wide range of ages. Thus effective dose applies to a population, not to
a specific individual and should not be used as the patients dose for the purpose of
assigning risk. This all too common misuse of effective dose, for a purpose for which
it was never intended and does not apply, is discussed in further detail in Chapters11
and 16. ICRPs initial recommendations for wT values (ICRP Publication 26, 1977)
were applied as shown in Equation 3-25, the product of which was referred to as the
effective dose equivalent (HE). Many regulatory agencies in the United States, including
the NRC, have not as yet adopted the current ICRP Publication 103 wT values. While
there is a current effort to update these regulations, they are, at present, still using the
old (1977) ICRP Publication 26 dosimetric quantities.
Summary
Most other countries and all the scientific literature use SI units exclusively. While
some legacy (traditional) radiation units and terms are still being used for regulatory
compliance purposes and some radiation detection instruments are still calibrated
in these units, students are encouraged to use the current SI units discussed above.
A summary of the SI units and their equivalents in traditional units is given in
Table3-6. A discussion of these radiation dose terms in the context of x-ray dosimetry
in projection imaging and CT is presented in Chapter 11.
SUGGESTED READING
Bushberg JT. The AAPM/RSNA physics tutorial for residents. X-ray interactions. RadioGraphics
1998;18:457468.
International Commission on Radiation Units and Measurements. Fundamental quantities and units for
ionizing radiation. Journal of the ICRU Vol 11 No 1 (2011) Report 85.
International Commission on Radiological Protection, Recommendations of the ICRP. Annals of the ICRP
37(24), Publication 103, 2007.
Johns HE, Cunningham JR. The physics of radiology, 4th ed. Springfield, IL: Charles C Thomas, 1983.
McKetty MH. The AAPM/ RSNA physics tutorial for residents. X-ray attenuation. RadioGraphics 1998;
18:151163.
The total harm to health experienced by an exposed group and its decendants as a result of
the groups exposure to a radiation source. Detriment is a multimentional concept. Its principal
components are the stochastic quantities: probability of attributable fatal cancer, weighted
probability of attributable non-fatal cancer, weighted probability of severe heritable effects, and
length of life lost if the harm occurs.
1
Chapter
Image Quality
Unlike snapshots taken on the ubiquitous digital camera, medical images are
acquired not for aesthetic purposes but out of medical necessity. The image quality
on a medical image is related not to how pretty it looks but rather to how well it conveys anatomical or functional information to the interpreting physician such that an
accurate diagnosis can be made. Indeed, radiological images acquired with ionizing
radiation can almost always be made much prettier simply by turning up the radiation levels used, but the radiation dose to the patient then becomes an important
concern. Diagnostic medical images therefore require a number of important tradeoffs in which image quality is not necessarily maximized but rather is optimized to
perform the specific diagnostic task for which the exam was ordered.
The following discussion of image quality is also meant to familiarize the reader
with the terms that describe it, and thus the vernacular introduced here is important as
well. This chapter, more than most in this book, includes some mathematical discussion that is relevant to the topic of image science. To physicians in training, the details of
the mathematics should not be considered an impediment to understanding but rather
as a general illustration of the methods. To image scientists in training, the mathematics
in this chapter are a necessary basic look at the essentials of imaging system analysis.
4.1
Spatial Resolution
Spatial resolution describes the level of detail that can be seen on an image. In
simple terms, the spatial resolution relates to how small an object can be seen on a
particular imaging systemand this would be the limiting spatial resolution. However, robust methods used to describe the spatial resolution for an imaging system
provide a measure of how well the imaging system performs over a continuous range
of object dimensions. Spatial resolution measurements are generally performed at
high dose levels in x-ray and g-ray imaging systems, so that a precise (low noise)
assessment can be made. The vast majority of imaging systems in radiology are digital, and clearly the size of the picture element (pixel) in an image sets a limit on what
can theoretically be resolved in that image. While it is true that one cannot resolve an
object that is smaller than the pixel size, it is also true that one may be able to detect
a high-contrast object that is smaller than the pixel size if its signal amplitude is large
enough to significantly affect the gray scale value of that pixel. It is also true that
while images with small pixels have the potential to deliver high spatial resolution,
many other factors also affect spatial resolution, and in many cases, it is not the pixel
size that is the limiting factor in spatial resolution.
The Spatial Domain

In radiology, images vary in size from small spot images acquired in mammography
(~50 mm50mm) to the chest radiograph, which is 350 mm430 mm. These images
60
Chapter 4 Image Quality
Input Point Stimulus
61
Output: PSF(x, y)
FIGURE 4-1 A point stimulus to an imaging system is illustrated (left), and the response of the imaging
system, the point spread function (PSF) is shown (right). This PSF is rotationally symmetric.
are acquired and viewed in the spatial domain. The spatial domain refers to the two
dimensions of a single image, or to the three dimensions of a set of tomographic
images such as computed tomography (CT) or magnetic resonance imaging (MRI).
A number of metrics that are measured in the spatial domain and that describe the
spatial resolution of an imaging system are discussed below.
The Point Spread Function, PSF

The point spread function (PSF) is the most basic measure of the resolution properties of an imaging system, and it is perhaps the most intuitive as well. A point source
is input to the imaging system, and the PSF is (by definition) the response of the
imaging system to that point input (Fig. 4-1). The PSF is also called the impulse
response function. The PSF is a two-dimensional (2D) function, typically described
in the x and y dimensions of a 2D image, PSF(x,y). Note that the PSF can be rotationally symmetric, or not, and an asymmetrical PSF is illustrated in Figure 4-2. The
diameter of the point input should theoretically be infinitely small, but practically
speaking, the diameter of the point input should be five to ten times smaller than the
width of the detector element in the imaging system being evaluated.
To produce a point input on a planar imaging system such as in digital radiography or fluoroscopy, a sheet of attenuating metal such as lead, with a very small hole
in it*, is placed covering the detector, and x-rays are produced. High exposure levels
need to be used to deliver a measurable signal, given the tiny hole. For a tomographic
FIGURE 4-2 A PSF which
is not rotationally symmetric is shown.
*In reality, PSF and other resolution test objects are precision-machined tools and can cost over a thousand dollars.
62
Stationary Imaging System
Non-stationary Imaging System

FIGURE 4-3 A stationary or shift-invariant imaging system is one in which the PSF remains constant over
the field of view of the imaging system. A nonstationary system has a different PSF, depending on the location
in the field of view.
system, a small-diameter wire or fiber can be imaged with the wire placed normal to
the tomographic plane to be acquired.
An imaging system with the same PSF at all locations in the field of view is called
stationary or shift invariant, while a system that has PSFs that vary depending on the
position in the field of view is called nonstationary (Fig. 4-3). In general, medical
imaging systems are considered stationaryeven if some small nonstationary effects
are present. Pixelated digital imaging systems have finite detector elements (dexels),
commonly square in shape (but not always), and in some cases, the detector element
is uniformly sensitive to the signal energy across its surface. This implies that if there
are no other factors that degrade spatial resolution, the digital sampling matrix will
impose a PSF, which is square in shape (Fig. 4-4) and where the dimensions of the
square are the dimensions of the dexels.
The PSF describes the extent of blurring that is introduced by an imaging system,
and this blurring is the manifestation of physical events during the image acquisition
FIGURE 4-4 A 2D RECT function is shown, illustrating the PSF of a digital imaging system with square
detector elements of width a, in the absence of any other blurring phenomenon. This PSF is the best possible
for a digital imaging system.
63
or reconstruction process. Mathematically, for a linear stationary system, the input

image is convolved with the PSF and that results in the output image. Convolution is
a mathematical operation discussed later in this chapter.
The Line Spread Function, LSF

When an imaging system is stimulated with a signal in the form of a line, the line
spread function (LSF) can be evaluated. For a planar imaging system, a slit in some
attenuating material can be imaged and would result in a line on the image. For a
tomographic imaging system, a thin plane of material can be imaged normal to the
tomographic plane, and that also would result in a line being produced in the image
(Fig. 4-5). Once the line is produced on the image (e.g., parallel to the y-axis), a
profile through that line is then measured perpendicular to the line (i.e., along the
x-axis). The profile is a measure of gray scale as a function of position. Once this
profile is normalized such that the area is unity, it becomes the LSF(x). As will be
discussed later, the LSF has an important role in the practical assessment of spatial
resolution.
For an analog imaging system such as radiographic film, a device called a scanning
microdensitometer is required to measure the optical density as a function of position.
The optical density trace would then have to be corrected for its nonlinear relationship
with incident exposure, before the LSF can be computed. For digital x-ray imaging
systems that have a linear response to x-rays such as digital radiography or CT, the
profile can be easily determined from the digital image using appropriate software.
Profiles from ultrasound and MR images can also be used to compute the LSF, as long
as the signal is converted in a linear fashion into gray scale on the images.
Slit imaging in projection radiography
Imaging a plane in tomographic systems
Image of Line
Profile of line is
measured perpendicular
to the line
FIGURE 4-5 Methods for acquiring the LSF are shown. The LSF can be acquired by imaging a slit image
(A), or by imaging a thin plane of material in a tomographic imaging system such as CT, MRI, or ultrasound (B).
C. The LSF is computed by taking an orthogonal profile across the measured slit image.
64
110
100
90
80
70
60
50
40
30
20
10
0
Amplitude (A.U.)
Amplitude (A.U.)
Position (mm)
110
100
90
80
70
60
50
40
30
20
10
0
2
Position (mm)
FIGURE 4-6 A. The conventional approach to measuring the LSF involves the acquisition of a slit image
along the y-axis, and a profile (gray scale versus position) across it (in the x-axis) is evaluated. This approach is
legitimate but is limited by the sampling pitch of the detector. B. An alternative approach to computing the LSF
was proposed by Fujita. Here, the slit image is acquired at a slight angle (5 degrees) relative to the y-axis. At
any row along the vertical slit, a single LSF can be measured. However, because of the angle, a composite LSF
can be synthesized by combining the LSFs from a number of rows in the image. The angle of the slit creates a
small differential phase shift of the LSF from row to rowthis can be exploited to synthesize an LSF with much
better sampling than the pixel pitch. This oversampled LSF is called the presampled LSF.
For a system with good spatial resolution, the LSF will be quite thin, and thus
on the measured profile, the LSF will only be few pixels wide, and the LSF measurement will consequently suffer from coarse pixel sampling in the image. One way to
get around this is to place the slit at a small angle (e.g., 2 to 8 degrees) to the detector element columns during the physical measurement procedure. Then, instead of
taking one profile through the slit image, a number of profiles are taken at different
locations vertically (Fig. 4-6) and the data are synthesized into a presampled LSF. Due
to the angulation of the slit, the profiles taken at different vertical positions (Fig. 4-6)
intersect the slit image at different phases (horizontally) through the pixel width
this allows the LSF to be synthesized with sub-pixel spacing intervals, and a better
sampled LSF measurement results.
The Edge Spread Function, ESF

In some situations, PSF and LSF measurements are not ideally suited for a specific
imaging application, where the edge spread function (ESF) can be measured. Instead
of stimulating the imaging system with a slit image as with the LSF, a sharp edge is
presented. The edge gradient that results in the image can then be used to measure the
ESF(x). The ESF is particularly useful when the spatial distribution characteristics of
glare or scatter phenomenon are the subject of interestsince a large fraction of the
field of view is stimulated, low-amplitude effects such as glare or scatter (or both)
become appreciable enough in amplitude to be measurable. By comparison, the tiny
area of the detector receiving signal in the PSF or LSF measurements would not be
sufficient to cause enough optical glare or x-ray scatter to be measured. A sharp,
straight edge is also less expensive to manufacture than a point or slit phantom.
An example of spatial domain spread functions is shown in Figure 4-7.
Position (mm)
PSF(x, y)
65
Position (mm)
LSF(x)
ESF(x)
FIGURE 4-7 The three basic spread functions in the spatial domain are shownthe PSF(x,y) is a 2D spread
function. The LSF and ESF are both 1D spread functions. There is mathematical relationship between the three
spread functions, as discussed in the text.
4.2
Convolution
Convolution is an integral calculus procedure that accurately describes mathematically
what the blurring process does physically. The convolution process is also an important
mathematical component of image reconstruction, and understanding the basics of
convolution is essential to a complete understanding of imaging systems. Below, a basic
description of the convolution process is provided. Convolution in 1D is given by
G ( x ) H ( x ') k ( x x ') dx ' H ( x ) k ( x )
[4-1]
where is the mathematical symbol for convolution. Convolution can occur in two
or three dimensions as well, and the extension to multidimensional convolution is
straightforward. Referring to Figure 4-8, a column of numbers H is to be convolved
with the convolution kernel, resulting in column G. The function H can also be thought
H
42.4
87.8
36.2
71.5
76.0
20.7
83.4
75.5
62.2
Kernel
0.20
0.20
0.20
0.20
0.20
62.8
58.5
57.6
65.4
63.6
42.4
87.8
36.2
71.5
76.0
20.7
83.4
75.5
62.2
Kernel
0.20
0.20
0.20
0.20
0.20
62.8
58.5
57.6
65.4
63.6
H
42.4
87.8
36.2
71.5
76.0
20.7
83.4
75.5
62.2
Kernel
0.20
0.20
0.20
0.20
0.20
62.8
58.5
57.6
65.4
63.6
FIGURE 4-8 The basic operation of discrete convolution is illustrated in this figure. In the three panes (AC),
the function H is convolved with the convolution kernel, resulting in the function G. The process proceeds by
indexing one data element in the array, and this is shown in the three different panes. The entire convolution
is performed over the entire length of the input array H.
66

100
90
H(x)
80
Gray Scale
70
G(x)
60
50
40
30
20
10
0
0
10
12
14
Pixel Number
16
18
20
22
FIGURE 4-9 A plot of the noisy input data H(x), and of the smoothed function G(x), is shown. The first elements in the data shown on this plot correspond to the data illustrated in Figure 4-8. The input function H(x)
is simply random noise. The convolution process with the RECT function results in substantial smoothing, and
this is evident by the much smoother G(x) function in comparison to H(x).
of as a row of pixel data on a 1D image. The kernel in this example is five elements
long, and its value sums to unity. In pane A, the first five numbers of H are multiplied
by the corresponding elements in the kernel (here, each element of the kernel has a
numerical value of 0.20); these five products are summed, resulting in the first entry
in column G. In pane B, the convolution algorithm proceeds by shifting down one
element in column H, and the same operation is computed, resulting in the second
entry in column G. In pane C, the kernel shifts down another element in the array of
numbers, and the same process is followed. This shift, multiply, and add procedure is
the discrete implementation of convolutionit is the way this operation is performed
in a computer. A plot of the values in columns H and G is shown in Figure 4-9, and
H in this case is randomly generated noise. The values of G are smoothed relative
to H, and that is because the elements of the convolution kernel in this case are
designed to perform data smoothingessentially by averaging five adjacent values in
Column H. A plot of various five-element kernels is shown in Figure 4-10. The
0.70
0.60
C
Kernel value
0.50
B
0.40
0.30
0.20
0.10
0.00
0.10
0.20
5
Pixel Number
FIGURE 4-10 Three convolution kernels are illustrated. A is a RECT function defined by [0.2, 0.2, 0.2, 0.2,
0.2], and the convolution process with this function is illustrated in Figures 4-8. Kernel B illustrates another
possible smoothing kernel, with values [0.05, 0.20, 0.50, 0.20, 0.05]. Kernel C has the values [0.15, 0.35,
0.50, 0.35, 0.15]. The negative sidebands of this kernel cause edge enhancement during a convolution procedure, but this can also increase the noise in the resulting function.
67
kernel illustrated in Figure 4-8 is kernel A in Figure 4-10, and the shape is that of
a rectangle (a RECT function), because all of the values in the kernel are the same.
Use of such a kernel in data analysis is also called a boxcar average or running average, with applications in economics and elsewhere. Kernel B in Figure 4-10 has an
appearance a bit similar to a bell-shaped curve, to the extent that only five values can
describe such a shape. Both kernels A and B have all positive values, and therefore,
they will always result in data smoothing. Kernel C has some negative edges on it,
and when negative values are found in a kernel, some edge enhancement will result.
In Equation 4-1 or Figure 4-8, it can be seen that negative values in the kernel will
result in weighted subtraction of adjacent elements of the input function, and this
brings out edges or discontinuities in the input function, and it tends to exacerbate
noise in the data as well. It is worth noting that if the kernel was [0, 0, 1, 0, 0], G
would be equal to H. This kernel is called a delta function.
Convolution can use kernels of any length, and for image processing, kernels from
3 elements to 511 or more elements are used. Although not required, most kernels
that are smaller in length are odd, so that there is a center element. For a kernel of
length Nk (odd), convolution is ill defined at the edges of the image for (Nk1)/2 pixels
(Fig. 4-8); however, assuming that the image has zero-valued pixels adjacent to its
edges is a common method for dealing with this. Above, only 1D convolution using 1D
kernels was discussed, but convolution is routinely performed in two dimensions using
2D kernels. In such case, 3 3 3, 5 3 5, and other odd kernel dimensions are routinely
used. Examples of 2D convolution kernels are illustrated in Figure 4-11. Three-dimensional convolution techniques are also used in medical imaging processing. Convolution is a mathematical process that describes physical blurring phenomena, but
convolution techniques can be used to restore (improve) spatial resolution as wellin
such cases the process is called deconvolution. While deconvolution can improve spatial
resolution in some cases, it also amplifies the noise levels in the image.
Relationships Between Spread Functions

There are mathematical relationships between the spatial domain spread functions PSF, LSF, and ESF. The specific mathematical relationships are described
here, but the take-home message is that under most circumstances, given a
gaussian blur
original
increasing edge enhancement
FIGURE 4-11 This figure illustrates the results of 2D image processing using a variety of different convolution kernels. The original image is shown as image B. Image A is a smoothed version, while images CE show
the effect of edge enhancement kernels with increasing edge enhancement from images C to E.
68
measurement of one of the spread functions, the others can be computed. This
feature adds to the flexibility of using the most appropriate measurement method
for characterizing an imaging system. These statements presume a rotationally
symmetric PSF, but in general, that is the case in many (but not all) medical
imaging systems.
The LSF and ESF are related to the PSF by the convolution equation. The LSF can
be computed by convolving the PSF with a line:
LSF ( x ) PSF ( x , y ) LINE ( y )
[4-2]
Because a line is purely a 1D function, the convolution shown in Equation 4-2

reduces to a simple integral
LSF ( x )
PSF ( x , y ) dy
y =
[4-3]
Convolution of the PSF with an edge results in the ESF
ESF ( x ) PSF ( x , y ) EDGE ( y )
[4-4]
In addition to the above relationships, the LSF and ESF are related as well. The ESF
is the integral of the LSF, and this implies that the LSF is the derivative of the ESF.

ESF ( x )
x '
LSF ( x ') dx '
[4-5]
One can also compute the LSF from the ESF, and the PSF can be computed from
the LSF; however, the assumption of rotational symmetry with the PSF is necessary
in the latter case.
4.3
Physical Mechanisms of Blurring

There are many different mechanisms in medical imaging that cause blurring, and
some of these mechanisms will be discussed in length in subsequent chapters. The
spatial resolution of an image produced by any optical device (such as in an image
intensifier/TV-based fluoroscopy system) can be reduced by accidental defocusing.
When an x-ray strikes an intensifying screen or other phosphor, it produces a burst
of light photons that propagate by optical diffusion though the screen matrix. For
thicker screens, the diffusion path toward the screens surface is longer and more
lateral diffusion occurs, which results in a broader spot of light reaching the surface
of the screen and consequently more blurring. Digital sampling usually results in the
integration of the 2D signal over the surface of the detector element, and this occurs
in digital x-ray, CT, nuclear medicine, ultrasound, and other modalities. Figure 4-4
illustrates the 2D RECT function that represents digital sampling, and the spatial
resolution decreases as the width of the RECT function increases. These are just some
examples of physical factors that reduce spatial resolution; many other processes can
degrade resolution as well.
The measurement of the various spread functions discussed above is used to characterize the impact of physical blurring phenomena in medical imaging systems. Subsequent
analyses on spread functions, and in particular on the LSF, are often performed to characterize the frequency dependence of spatial resolution, as described below.
Line(y) 5 1, where y 5 0,
5 0, elsewhere
Edge(y) 5 1, where y > 0,

5 0, elsewhere
4.4
69
The Frequency Domain

The PSF, LSF, and ESF are apt descriptions of the resolution properties of an imaging
system in the spatial domain. Another useful way to express the resolution of an
imaging system is to make use of the spatial frequency domain. By analogy, the amplitude of sound waves vary as a function of time (measured in seconds), and temporal
frequency is measured in units of cycles/s (s21), also known as Hertz. For example,
the middle A key on a well-tuned piano corresponds to 440 Hz. If the period of a
sound wave occurs within shorter periods of time, the sound wave is of higher frequency. Spatial frequency is similar to temporal frequency except that distance (in
mm) is used instead of time. In the spatial frequency domain, smaller objects correspond to higher frequencies, and larger objects correspond to lower frequencies.
There are some technical nuances to the statement, but to a first approximation, spatial frequency can be thought of as just a different way of thinking about
object sizelow spatial frequencies correspond to larger objects in the image, and
higher spatial frequencies correspond to smaller objects. If the size of an object (D)
is known, it can be converted to spatial frequency (F51/2D), and if the spatial
frequency (F) is known, it can be converted to object size (D51/2F). Here, the
dimensions of D are mm, and the dimensions of F are cycles/mm, which is equivalent
to mm21 or 1/mm. These statements are true, but a rigorous mathematical description of the relationship between D and F would involve the shape, size, and periodicity of the object, but that discussion is not necessary here.
Fourier Series and the Fourier Transform

In Figure 4-12, let the solid black line be a trace of gray scale as a function of position
across an image. Nineteenth-century French mathematician Joseph Fourier developed a method for decomposing a function such as this gray scale profile into the
sum of a number of sine waves. Each sine wave has three parameters that characterize its shape: amplitude (a), frequency (f), and phase (), where
g ( x ) a sin(2 fx + y )
[4-6]
Figure 4-12 illustrates the sum of four different sine waves (solid black line),
which approximates the shape of two rectangular functions (dashed lines). Only four
300
Gray Scale
250
200
150
100
50
0
0
10 12 14 16
Position (mm)
18
20
22
24
FIGURE 4-12 The basic concept of Fourier Series analysis is illustrated. The solid black lines are the sum of the
four sets of sine waves, and approximate the two RECT functions (dashed lines). Only four sine waves are summed
here, but with a more complete set, the RECT functions could be almost perfectly matched. The Fourier transform
breaks any arbitrary signal down into the sum of a set of sine waves of different phase, frequency, and amplitude.
70
sine waves were used in this figure for clarity; however, if more sine waves (a more
complete Fourier series) were used, the shape of the two rectangular functions could
in principle be matched.
The Fourier transform is an algorithm that decomposes a spatial or time domain
signal into a series of sine waves that, when summed, replicate that signal. Once a
spatial domain signal is Fourier transformed, the resulting data are considered to be
in the frequency domain. The Fourier transform is given by
G( f )
g ( x ) e 2ifx dx
[4-7]
where i is 1 . If, for example, the input signal g(x) to the Fourier transform is a
sine wave with a frequency , the function G(f) will be a plot with spikes at frequencies 6, and it will be zero elsewhere. The function G(f) will in general consist of
complex numbers, with real and imaginary components.
The Fourier transform, FT[], converts a temporal or spatial signal into the frequency domain, while the inverse Fourier transform, FT21[], converts the frequency
domain signal back to the temporal or spatial domain. The Fourier transform can be
used to perform convolution. Referring back to Equation 4-1, the function G(x) can
be alternatively computed as

G ( x ) FT 1 {FT [H ( x )] FT [k ( x )]}
[4-8]
Equation 4-8 computes the same function G(x) as in Equation 4-1, but
c ompared to the convolution procedure, in most cases the Fourier transform computation runs faster on a computer. Therefore, image processing methods that
employ convolution filtration procedures such as in CT are often performed in
the frequency domain for computational speed. Indeed, the kernel used in CT is
often described in the frequency domain (descriptors such as ramp, Shepp-Logan,
bone kernel, B41), so it is more common to discuss the shape of the kernel in the
frequency domain (e.g., FT[k(x)]) rather than in the spatial domain (i.e., k(x)), in
the parlance of clinical CT.
Fourier computations are a routine part of medical imaging systems. Fourier
transforms are used to perform the filtering procedure in filtered back projection,
for CT reconstruction. Inverse Fourier transforms are used in MRI to convert the
measured time domain signal into a spatial signal. Fourier transforms are also used
in ultrasound imaging and Doppler systems.
The Modulation Transfer Function, MTF(f)

Conceptual Description
Imagine that it is possible to stimulate an imaging system spatially with a pure sinusoidal wave form, as illustrated in Figure 4-13A. The system will detect the incoming
sinusoidal signal at frequency f, and as long as the frequency does not exceed the
Nyquist frequency (discussed later) of the imaging system (i.e., f , FN), it will produce an image at that same frequency but in most cases with reduced contrast (Fig.
4-13A, right side). The reduction in contrast transfer is the result of resolution losses
in the imaging system. For the input sinusoidal signals with frequencies of 1, 2, and
4 cycles/mm (shown in Fig 4-13A), the recorded contrast levels were 87%, 56%,
and13%, respectively, on the measured images. For any one of these three frequencies
measured individually, if the Fourier transform was computed on the recorded signal,
the result would be a peak at the corresponding frequency (Fig. 4-13A, right side).
Three such peaks are shown in Figure 4-13B, representing three sequentially and
71
individually acquired (and then Fourier transformed) signals. The amplitude of the
peak at each frequency reflects the contrast transfer (retained) at that frequency, with
contrast losses due to resolution limitations in the system. Interestingly, due to the
characteristics of the Fourier transform, the three sinusoidal input waves shown
in Figure 4-13A could be acquired simultaneously by the detector system, and the
Input signal
Measured signal
1 cy / mm
87 %
2 cy / mm
56 %
4 cy / mm
13 %
1 mm
A
1.0
0.9
0.8
MTF (f)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
2
3
Spatial Frequency (cy / mm)
B
FIGURE 4-13 A. Sinusoidal input signals are incident on a detector (intensity as a function of position), and
three different frequencies are shown as input functions (left). The signals measured by the imaging system are
shown on the rightthe frequency is the same as the inputs in all cases, but the amplitude of the measured signal is reduced compared to that of the input signal. This reduction in amplitude is a result of resolution losses in
the imaging system, which are greater with signals of higher frequencies. For the input at 1 cycle/mm, the original
100% amplitude was attenuated to 87%, and with the 2- and 4-cycle/mm input functions, the resulting signal
amplitudes were reduced to 56% and 13%, respectively. B. This figure shows the amplitude reduction as a function of spatial frequency shown in A. At 1 cycle/mm, the system reduced the contrast to 87%of the input. For
2 mm21 and 4 mm21, the signal was modulated as shown. This plot shows the MTF, which illustrates the spatial
resolution of an imaging system as a function of the spatial frequency of the input signal.
72
1.0
0.9
0.8
MTF(f)
0.7
0.6
0.5
limiting spatial resolution
0.4
0.3
0.2
10 % MTF
0.1
0.0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Spatial Frequency (cycles / mm)

FIGURE 4-14 The limiting spatial resolution is the spatial frequency at which the amplitude of the MTF
decreases to some agreed-upon level. Here the limiting spatial resolution is shown at 10% modulation, and
the limiting spatial resolution is 1.6 cycles/mm.
Fourier Transform could separate the individual frequencies and produce the three
peaks at F 5 1, 2, and 4 cycles/mm shown in Figure 4-13B. Indeed, if an input
signal contained more numerous sinusoidal waves (10, 50, 100,.) than the three
shown in Figure 4-13A, the Fourier transform would still be able to separate these frequencies and convey their respective amplitudes from the recorded signal, ultimately
resulting in the full, smooth modulation transfer function (MTF) curve shown in
Figure4-13B.
Practical Measurement
It is not possible in general to stimulate a detector system with individual, spatial
sinusoidal signals as described in the paragraph above. Rather, the LSF, discussed
previously and illustrated in Figures 4-6 and 4-7, is used to determine the MTF in
experimental settings. A perfect line source input (called a delta-function), it turns
out, is represented in the frequency domain by an infinite number of sinusoidal
functions spanning the frequency spectrum. Therefore, the Fourier transform of the
LSF(x) computes the full MTF curve as shown in Figure 4-13B. Prior to the computation, the LSF(x) is normalized to have unity area

LSF ( x ) dx 1
[4-9]
Then the Fourier transform is computed, and the modulus (brackets) is taken,
resulting in the MTF(f)

MTF ( f ) =
LSF ( x ) e 2ifx dx
[4-10]
Limiting Resolution
The MTF gives a rich description of spatial resolution, and is the accepted standard
for the rigorous characterization of spatial resolution. However, it is often useful
to have a single number value that characterizes the approximate resolution limit
of an imaging system. The limiting spatial resolution is often considered to be the
frequency at which the MTF crosses the 10% level (see Fig. 4-14), or some other
agreed-upon and specified level.

Sampling pitch
Aperture width
73
FIGURE 4-15 This figure illustrates a side view of detector elements in a hypothetical detector system.
Detector systems whose center-to-center spacing (pitch) is about equal to the detector width are very common, and these represent contiguous detector elements. The sampling pitch affects aliasing in the image,
while the aperture width of the detector element influences the spatial resolution (the LSF and MTF). A sine
wave (shown) where each period just matches the width of two detector elements is the highest frequency
sine wave that can be imaged with these detectors due to their sampling pitch.
Nyquist Frequency
Lets look at an example of an imaging system where the center-to-center spacing
between each detector element (dexel) is D in mm. In the corresponding image, it
would take two adjacent pixels to display a full cycle of a sine wave (Fig. 4-15)one
pixel for the upward lobe of the sine wave, and the other for the downward lobe. This
sine wave is the highest frequency that can be accurately measured on the imaging
system. The period of this sine wave is 2D, and the corresponding spatial frequency
is FN 5 1/2D. This frequency is called the Nyquist frequency (FN), and it sets the
upper bound on the spatial frequency that can be detected for a digital detector
system with detector pitch D. For example, for D 5 0.05 mm, FN 5 10 cycles/mm,
and for D 5 0.25 mm, FN 5 2.0 cycles/mm.
If a sinusoidal signal greater than the Nyquist frequency were to be incident upon
the detector system, its true frequency would not be recorded, but rather it would
be aliased. Aliasing occurs when frequencies higher than the Nyquist frequency are
imaged (Fig. 4-16). The frequency that is recorded is lower than the incident frequency, and indeed the recorded frequency wraps around the Nyquist frequency. For
example, for a system with D 5 0.100 mm, and thus FN 5 5.0 cycles/mm, sinusoidal inputs of 2, 3, and 4 cycles/mm are recorded accurately because they obey the
Nyquist Criterion. The Fourier transform of the detected signals would result in a
measured
signal
input signal
digital sampling
FIGURE 4-16 The concept of aliasing is illustrated. The input sine wave is sampled with the sampling comb
(arrows), but the Nyquist criterion is violated here because the input sine wave is higher than the Nyquist frequency. Thus, the sampled image data will be aliased and the lower frequency sine wave will be seen as the
measured signal on the image.
Relative amplitude
74
Fin = 2
Fin = 6
Fin = 3
Fin = 7
Fin = 4
Fin = 8
3
4
5 0
1
2
3
Spatial Frequency (cycles/mm)
FIGURE 4-17 For a single-frequency sinusoidal input function to an imaging system (sinusoidally varying
intensity versus position), the Fourier transform of the image results in a spike (delta function) indicating the
measured frequency. The Nyquist frequency in this example is 5 cycles/mm. For the three input frequencies in
the left panel, all are below the Nyquist frequency and obey the Nyquist criterion, and the measured (recorded)
frequencies are exactly what was input into the imaging system. On the right panel, the input frequencies
were higher than the Nyquist frequency, and the recorded frequencies in all cases were aliasedthey wrapped
around the Nyquist frequencythat is, the measured frequencies were lower than the Nyquist frequency by
the same amount by which the input frequencies exceeded the Nyquist frequency.
histogram of the frequency distribution of the measured sinusoid frequencies, but

since the input functions are single frequencies, the Fourier transform appears as a
spike at that frequency (Fig. 4-17). For sinusoidal inputs with frequencies greater
than the Nyquist frequency, the signal wraps around the Nyquist frequencya frequency of 6 cycles/mm (FN 1 1) is recorded as 4 cycles/mm (FN21), and a frequency
of 7 cycles/mm (FN 1 2) is recorded as 3 cycles/mm (FN22), and so on. This is seen
on Figure 4-17 as well. Aliasing is visible when there is a periodic pattern that is
imaged, such as an x-ray antiscatter grid, and aliasing appears visually in many cases
as a Moir pattern or wavy lines.
The Presampled MTF

Aliasing can pose limitations on the measurement of the MTF, and the finite size of
the pixels in an image can cause sampling problems with the LSF that is measured
to compute the MTF (Equation 4-10). Figure 4-6 shows the angled slit method for
determining the presampled LSF, and this provides a methodology for computing
the so-called presampled MTF. Using a single line perpendicular to the slit image
(Fig. 4-6A), the sampling of the LSF measurements is D, and the maximum frequency that can be computed for the MTF is then 1/2D. However, for many medical
imaging systems it is both possible and probable that the MTF has nonzero amplitude beyond the Nyquist limit of FN 5 1/2D. In order to measure the presampled
MTF, the angled-slit method is used to synthesize the presampled LSF. By using a
slight angle between the long axis of the slit and the columns of detector elements in
the physical measurement of the LSF, different (nearly) normal lines can be sampled
(Fig. 4-6B). The LSF computed from each individual line is limited by the D sampling pitch, but multiple lines of data can be used to synthesize an LSF, which has
much better sampling than D. Indeed, oversampling can be by a factor of 5 or 10,
depending on the measurement procedure and how long the slit is. The details of
how the presampled LSF is computed are beyond the scope of the current discussion; however, by decreasing the sampling pitch from D to, for example, D/5, the
75
Nyquist limit goes from FN to 5FN, which is sufficient to accurately measure the
presampled MTF.
RECT Functions and SINC
We have seen in the above discussion on aliasing, that the discrete sampling in
digital detector systems imposes the Nyquist Criterion (i.e., FN 5 1/2D) and if
this criterion is not met, then aliasing may occur. Digital detector systems such
as flat panel detector systems (as a concrete example) have detector elements that
are essentially contiguous (i.e., very little dead space between adjacent detector
elements). Thus, for a detector where the detector pitch (the center-to-center distance between adjacent detector elements) is D, the width of each detector element
(a) is about the same (Fig. 4-16), that is, a D. This finite width of each detector
element means that all of the signal which is incident on each dexel is essentially
averaged together, and one number (the gray scale value) is produced for each
detector element (in general). This means that on top of any blurring that may
occur from other sources, the width of the dexel imposes a fundamental limit on
spatial resolution. In a very literal sense, the detector widthidealized as a rectangular (RECT) response functionis the best the LSF can ever get on a digital imaging system. The RECT function is a rectangle-shaped function of width a, usually
centered at x5 0, such that the RECT function runs from 2a/2 to 1a/2. Let the
LSF be a RECT function with width a, and the MTF is then computed as the Fourier
transform of the LSF. This Fourier transform can be computed analytically, and the
resulting MTF will be given by
x
MTF ( f )
RECT e 2ifx dx a sinc(af )
[4-11]
x
a
A plot of this MTF is illustrated in Figure 4-18. The function on the right side of
Equation 4-11 is commonly called a sinc function. The sinc function goes to zero at
F 5 1/a, but aliasing occurs above the Nyquist frequency at 1/2D. Thus, for a system
which has an MTF defined by the sinc function, the possibility of aliasing is high
1.0
FN =
0.9
0.8
1
2
1
a
sinc(f)
MTF(f)
0.7
0.6
a
0.5
0.4
0.3
0.2
0.1
0.0
0
4
5
6
7
8
3
Spatial Frequency (cycles / mm)
10
11
FIGURE 4-18 This figure shows the MTF for a RECT input function of width a. Here, a = 0.100 mm.
This was a contiguous detector; so the sampling pitch, D, was equal to a. Thus, the Nyquist frequency is
5 cycles/mm and the amplitude of the MTF goes to zero at 1/a = 10 cycles/mm. This MTF shows the SINC
functionthe analytical Fourier transform of the RECT function of width a. For an imaging system with a
detector aperture of width a, this curve represents the best MTF possible for that detector.
76

FIGURE 4-19 Line pair phantoms are used for
practical estimates of the spatial resolution in field
tests of clinical imaging equipment. An ideal test
phantom (left) is imaged, and the system blurs the
image of the object as shown here (right). The
observer reads the spatial frequency (not indicated
here) corresponding to the smallest set of bars that
were visible, and this measurement is considered
the limiting spatial resolution.
since the MTF still has nonzero amplitude above FN. An imaging system that has an
MTF defined by the sinc function is performing at its theoretical maximum in terms
of spatial resolution.
Field Measurements of Resolution Using Resolution Templates

Spatial resolution should be monitored on a routine basis for many imaging modalities. However, measuring the LSF or the MTF is more detailed than necessary for
routine quality assurance purposes. For most clinical imaging systems, the evaluation of spatial resolution using resolution test phantoms is adequate for routine
quality assurance purposes. The test phantoms are usually line pair phantoms or
star patterns (Fig. 4-19). The test phantoms are imaged, and the images are viewed
to estimate the limiting spatial resolution of the imaging system. There is a degree of
subjectivity in such an evaluation, but in general viewers will agree within acceptable
limits. These measurements are routinely performed on fluoroscopic equipment,
radiographic systems, nuclear cameras, and in CT.
4.5
Contrast Resolution
Contrast resolution refers to the ability to detect very subtle changes in gray scale
and distinguish them from the noise in the image. Contrast resolution is characterized by measurements that pertain to the signal-to-noise ratio (SNR) in an image.
Contrast resolution is not a concept that is focused on physically small objects
per se (that is the concept of spatial resolution); rather, contrast resolution relates
more to anatomical structures that produce small changes in signal intensity (image
gray scale), which make it difficult for the radiologist to pick out (detect) that structure from a noisy background.
Basic Notions of Accuracy and Precision

Accuracy and precision are illustrated using targets in Figure 4-20. Accuracy
relates to how close one gets to the truth (the bulls-eye, in this case), while
precision is a description of the variation, scatter, or reproducibility in a measurement. Measurements (or marksmanship) with low precision have higher levels of noise. Improving precision in a measurement may require new equipment
(a better target rifle) or more practice in the case of the marksman, and once the
A.
Accurate but not precise
C. Precise but not accurate
77
B. Neither accurate nor precise
D.
Accurate and precise
FIGURE 4-20 Basic concepts of accuracy and precision are illustrated.
marksman becomes more reproducible, he or she can make adjustments to correct

for bias that may affect accuracy. In Figure 4-20C, a tight pattern is achieved (high
precision) but there is a bias to the upper right of the bulls-eye. Correction for
this bias may require adjustments to the rifles sites, and after such adjustments
(Fig. 4-20D), accuracy with precision is achieved. This is an example where the
bias can be corrected for.
In medical images, precision has to do with the amount of noise in the image. In
the case of x-ray and gamma ray imaging, which are typically quantum limited, precision can almost always be improved by collecting more photons (quanta) to make
the imageby increasing the time of the image acquisition, turning up the intensity
of the source (or injecting more isotope), or both. In MRI, a number of redundant
signal measurements are made and are averaged together to improve precision and
reduce noise.
Sources of Image Noise

Grain Noise
When radiographic film is exposed and processed, the increase in observed darkness (optical density) in the film corresponds to an increase in the density of
reduced silver grains adhering to the film support. The actual distribution of
the silver grains is random and is a function of the manufacturing and development processes. Normally, the individual grains are not resolvable by the unaided
human eye, and thus grain noise is not a factor when reading a film radiograph on
a view box. However, in screen-film mammography, it is the standard of practice
that the breast imager uses a magnifying glass to scan for small microcalcifications
on the image. Depending on the magnification factor, it is theoretically possible to
observe the silver grains, and thus, they would become a source of image noise in
that instance (Fig. 4-21A).
78
actual signal
signal + noise
electronic noise
A. Grain Noise
C. Structured Noise
B. Electronic Noise
D. Anatomical Noise
FIGURE 4-21 Various sources of image noise are illustrated. A. Grain noise results in screen film radiography, where the image is formed by millions of silver grains attached to the clear film substrate. In general,
grain noise is too small to be resolved in general radiography. B. Electronic noise is typically additive noise
from various sources that does not scale with the signal level. For the signal + noise curve in the figure,
most of the measured information is noise. This plot illustrates that additive noise is worst in areas of low
actual signal in the detector. C. Structured noise represents a reproducible pattern on the image that reflects
differences in the gain of individual detector elements or groups of detector elements. Structured noise can
be corrected using a flat field algorithm. D. Anatomical noise refers to anatomy in the patient which is not
pertinent to the specific imaging examination. For example, ribs are a source of anatomical noise in a chest
radiograph, where the lung parenchyma is the anatomy of interest. Ribs can be removed by using dual-energy
x-ray imaging techniques.
Electronic Noise
Electronic detector systems can be analog or digital, but the flow of electrons (current)
conveys the signal at one or more points in the imaging chain. There are electrons that
resulted from actual signal detection events (in CT, MRI, nuclear imaging, etc.), and there
are also electrons that are added to the signal that are not the result of signal detection
eventsthis added electronic noise can be from thermal noise, shot noise, and other
electronic noise sources. If the electrons associated with noise are added into the signal
prior to amplification circuits, then their noise contribution will be amplified as well. In
some imaging situations, especially when signal levels are low, the added noise can be
substantial and can contribute appreciably to the overall noise levels in the image (Fig.
4-21B). There are several ways to reduce electronic noise, including cooling the detector
system to reduce thermal noise, designing in noise reduction circuitry (e.g., double correlated sampling), or shielding electronics to avoid stray electronic signal induction.
Electronic noise is a real problem in some clinical applications, such as the use
of thin film transistor panels for fluoroscopy or cone beam CT, or in very low dose
CT situations.
79
Structure Noise
Most pixelated detectors have a number of parallel channels for reading out the
array of detector elements, and this reduces readout time. Each channel uses its
own amplifier circuits and these circuits cannot be perfectly tuned with respect
to each other. As a result, groups of detector elements that are read out may have
different offset noise and gain characteristics, and these cause structured or fixed
pattern noise in digital detector systems (Fig. 4-21C). The key to correcting for
structured noise is that it is spatially constant for a period of time. This allows the
offset and gain factors for each individual detector element to be characterized by
exposing the detector to radiation in the absence of an object (the so-called gain
image). The offset image is measured with no radiation incident on the detector.
These two calibration images can then be used to correct for structured noise using
the so-called flat field correction algorithm. In addition to a visible structure on
uncorrected images due to amplifier configuration, each individual detector element has its own gain and offset characteristics which are also corrected by the flat
field correction algorithm. Because it is probable that the structured noise pattern
can change over time, for many imaging systems it is routine to acquire the offset
and gain images necessary for calibration frequentlyhourly, daily, or monthly,
depending on the system.
There are other causes for structured noisea splash of iodinated contrast agent
on the input surface of an image intensifier (II) used in angiography will cause a dark
spot that is always at the same location in the image (except when magnification
modes are applied). Flat field correction approaches will correct for these blemishes
as well, but if the II is cleaned and the flat field images are not recomputed, then the
blemish will be overcompensated forit will be a bright spot.
Anatomical Noise
Anatomical noise is the pattern on the image that is generated by patient anatomy
that is always present but not important for the diagnosis. For example, in abdominal angiography, the vascular system is the anatomy of interest, and other sources of
image contrast such as bowel gas and spine just get in the way. Using digital subtraction angiography, images are acquired before and after the injection of vascular contrast agents, and these images are subtracted, revealing only the vascular anatomy.
Chest radiographs are acquired for the evaluation of the pulmonary anatomy and
mediastinum; the ribs are a source of anatomical noise. Using dual-energy subtraction techniques (Fig. 4-21D), the ribs (and other bones) can be subtracted revealing
only the soft tissue contrast in the image. Both temporal and dual-energy subtraction
methods are performed primarily to reduce anatomical noise.
One of the underappreciated aspects of modern tomographic imaging, such as
CT, MRI, and ultrasound, is that these images eliminate overlapping anatomical structures, and this substantially reduces anatomical noise by reducing or eliminating the
superposition of normal tissue structures. Much of the power of tomography, then, is
in its ability to reduce anatomical noise through spatial isolation. Thus, anatomical
noise can be reduced using a number of approaches that capitalize on the temporal,
energy, or spatial characteristics of the image formation process.
I corrected g
I raw I offset (r )
I gain I offset ( g )
, where g is the mean gray scale of the denominator, and the offsets are
for the raw (r) and gain (g) images, respectively.
80
Quantum Noise
The term quanta refers to any number of particles or objects that can be counted,
such as electrons, x-ray photons, optical photons, or even brush strokes on impressionist paintings. The modern human visual experience involves quantum-rich
viewinga sunny day, a well-illuminated room, a bright flat panel monitor. When
the number of quanta used to illuminate a scene is large, there is virtually no noise
in the visual image. In low-light imaging environments, such as when night vision
glasses are used, significant levels of optical quantum noise are seen. In the radiology department, to reduce radiation dose to the patient from x-rays or gamma rays,
imaging modalities based upon ionizing radiation use relatively few quanta to form
the imageindeed, the numbers of quanta are so low that for most medical images
involving x-rays or gamma rays, appreciable noise in the image results, and this noise
is called quantum noise. Before directly addressing the topic of imaging in a quantumlimited setting, it is useful to review the topic of statistical distributions.
Statistical Distributions
There is a large field devoted to the evaluation of random variation in data, and most
readers will likely have more than one book on statistics on their bookshelf. Here, a
basic discussion of statistical distributions is presented as a prelim to a more focused
discussion on contrast resolution in medical imaging.
Nonparametric Measures of Noise

Nonparametric metrics of a data distribution use measures that assume no underlying statistical distribution to the data distribution. The median is the point in the
distribution where 50% of the observations are greater, and 50% are less. It is the
halfway point in the observed data set. The mode is the most frequent observation,
the highest point in the histogram (Fig. 4-22). The width of the distribution is characterized by the range, and there is flexibility in the fractional range that one uses
the 100% range would be subject to outliers, and so often a smaller fraction of the
range is quoted, such as the 50% or 90% range.
Nonparametric characterizations of two distributions can be used to test for
differences, for example. Appropriate nonparametric statistical tests (Wilcoxon rank
sum, etc.) are used. The use of nonparametric statistical tests is applicable to any
distribution; however, there is greater power in tests of statistical inference when an
underlying parametric distribution (such as the normal distribution) can be assumed.
The normal distribution is discussed in the next section.
FIGURE 4-22 An arbitrary distribution is shown, with the nonparametric parameters of median,
mode, and range illustrated.
81

0.050
0.045
0.040
Frequency
0.035
0.030
0.025
0.020
0.015
0.010
0.005
0.000
0
10
20
30
40
50
60
Observation
70
80
90
100
FIGURE 4-23 Two normal distributions are illustrated. The normal distribution has two parameters, which
determine its shape, and position the mean x , and standard deviation s.
Most textbooks on statistical distributions start with a discussion of the binomial

distribution, which is a discrete distribution and is useful for discussing coin tosses
and the like. The normal distribution and the Poisson distributions are extensions of
the binomial distribution, and are the predominant statistical distributions used in
medical imaging. For that reason, we refer readers interested in the binomial distribution to any basic text on statistics.
The Normal Distribution

The normal distribution, also called the gaussian distribution, is the most widely
used statistical distribution in scientific analysis and other observational settings.
The well-known bell-shaped curve (Fig. 4-23) is characterized by two parametersthe mean (x) and the standard deviation (s). The square of the standard deviation is the variance, s2. On a 2D plot of the distribution, the x-axis
parameter typically is called the independent variable, and the y-axis parameter
is the dependent variable (y, G(x), etc). The normal distribution is given by the
equation

G(x ) =
1 x x

1
e 2
2
,
[4-12]
where x is the independent variable, x and s are the parameters of interest, and the
first term to the right of the equal sign simply normalizes the area of G(x) to unity.
Two normal distributions are illustrated in Figure 4-23, one with a mean of 40 and
the other with a mean of 60. It is clear from the graph that the mean describes the
center of the bell-shaped curve, and the value of s determines the width of each distribution. The value of s is the half width at 61% of full height, and that is useful for
estimating the value of s on generic bell-shaped plots. The mean can be computed
from a series of N observations xi, such as x1, x2, x3, xN, where

1
N
x .
i
i 1
[4-13]
82
The variance is computed as

N
(x x )
i1
N 1
,
[4-14]
and the standard deviation is the square root of the variance

=
2.
[4-15]
The dependence on two parameters (x and s) means that the normal distribution
is flexible and can be used to model a number of physical situations and laboratory measurements. Furthermore, the value of each parameter, x and s, can have a
physical interpretationnot just a statistical one. The following simple example will
illustrate this:
A large elementary school has a number of classes from kindergarten to the sixth
grade. The mean height of the children in the fourth grade (age 9) is 135 cm, while
the mean height of all the sixth graders (age 11) is 147 cm. It is not surprising
that grade-school children grow rapidly in height as they age, and thus the fact that
the mean height of sixth graders is greater than that of fourth graders is logical and
easy to understand. It has physical meaning. But what about the standard deviation?
This school is large and has 3 fourth grades (three classes of 30 students each) to
accommodate enrollment. While the average height of all the fourth graders in this
particular school is 135.7 cm (simulated data were used in this example), the standard deviation across all 90 students was 6.4 cm. Let us say that the students were
assigned randomly to each classroom, and then the mean and standard deviation
were measuredfor Class 4A it was 134.4 cm (s 5 6.4 cm), Class 4B was 137.7cm
(s 5 6.0 cm), and Class 4C was 135.3 cm (s 5 7.0 cm). Instead of randomly assigning children to each class, what if the fourth graders were assigned to each class by
heightline all 90 children up by height; the shortest 30 go to Class 4A, the tallest
30 go to Class 4C, and the rest are assigned to Class 4B (Fig. 4-24). In this case,
the mean and standard deviation for these classes are 128.2 cm (2.3 cm), 135.5 cm
class 4A
class 4B
class 4C
150
assignment by height
Height (cm)
145
140
135
130
125
random assignment
120
0
10
20
30
40
50
60
Student Number
70
80
90
FIGURE 4-24 The black line shows the heights of 90 children in fourth grade, and this is simply a random
order. The three sections show the heights of children in three different classrooms, where the children were
assigned to each classroom by height. The purpose of this illustration is to show the value of the two-parameter
normal distribution in describing physical observations.
83
(2.2cm), and 143.1 cm (1.9 cm). It is not surprising that, because the students in
this latter example were assigned by height, Class 4A is shorter, Class 4B is next,
and Class 4C has the largest mean height. Also, because the children were selected
by height, there is less variation in height in each class, and that is why the standard
deviation for each classroom for this latter assignment approach is substantially less
( 52.1) than random assignment ( 56.5). There is less variation in height in each
classroom, and thus the standard deviation measured in each class is lowerthis
makes physical sense.
In regard to the first randomization scheme mentioned in the above paragraph,
what methods would result in a true random assignment to each of the 3 fourthgrade classrooms? What if alphabetical order based on the last name of the child
was used, would this be random? Probably not, since last names can be related to
ethnicity, and there is a relationship between ethnicity and height. Such a scheme
would probably add bias to the assignment scheme that may not be recognized.
What if the childs birth date was used for the assignment? Those with birthdays in
December and January are bound to be older than those children with birthdays in
June and August, and since older children are in general taller (even just a little),
this randomization scheme would suffer from bias as well. This example is meant to
illustrate how unintentional bias can enter an experimentor, in radiology, selection criteria for a clinical trial. Perhaps the easiest way to eliminate bias in the above
example is to print the numbers 1 through 90 on identical chits, thoroughly mix
them up in a hat, and let each child pick from the hatNumbers 1 to 30 go to Class
4A, 31 to 60 to Class 4B, and 61 to 90 go to Class 4C.
Relative Areas of the Normal Distribution

With the normal distribution, the relative area of the curve as a function of 6s (in
both the negative and positive directions) is useful to know. Measurements that fall
within x 6 1s represent 68% of the observations that are being characterized by
the normal distribution (Fig. 4-25). Values falling within x 6 2s represent 95% of
Frequency (y)
0.050
0.045
0.040
0.035
0.030
0.025
0.020
0.015
0.010
0.005
0.000
0
10
20
30
40
50
60
70
80
90
100
Observation (x)
FIGURE 4-25 The areas of the normal distribution for x 6 n s, for n = 1, 2, and 3 are shown. For x 6 1s,
the area is 68%, for x 6 2s it is 95%, and for x 6 3s the area is 99% of the total area of the normal
distribution.
84
the observations, and values falling into the range x 6 3s encompass 99% of the
observations. In scientific papers, it is common to plot curves with error bars spanning
6 2s, and this demonstrates to the observer that 95% of expected observations for
each data point should fall between the ends of the error bars. A p-value of 0.05 is
commonly used as the threshold for statistical significance, and this states that there
is a 95% probability (120.05) that an observation that obeys the normal distribution
will fall inside the x 6 2s range, and a 5% probability (0.05) that it will fall outside
that range.
The Poisson Distribution

The normal distribution is a very commonly used statistical distribution; however,
it is not the only one. Another important statistical distribution that is relevant in
radiological imaging is the Poisson distribution, P(x), which is given by
P(x )
m x m
e
x!
[4-16]
where m is the mean and x is the independent variable. The very important thing
about the Poisson distribution is that its shape is governed by only one parameter
(m), not two parameters as we saw for the normal distribution. Figure 4-26 illustrates
the Poisson distribution with m 5 50 (dashed line), along with the normal distribution with x 550 and s 5 x . The normal distribution is often used to approximate
the Poisson distribution, in part because the factorial (x!, i.e., 5! 5 1 3 2333
4 3 5) in Equation 4-16 makes computation difficult for large values of x (e.g., 70!
. 10100). Therefore, we can approximate the Poisson distribution with the normal
distribution, but only when we stipulate that s 5 x . This is a very important stipulation, with very important ramifications in imaging.
X ray and g-ray counting statistics obey the Poisson distribution, and this
is quite fortunate for radiology. Why? We will use an x-ray imaging example
(i.e., radiography, mammography, fluoroscopy, CT), but this pertains to nuclear
0.06
Poisson Distribution (m=50)
0.05
Normal Distribution
(x=50, =50)
Frequency
0.04
0.03
0.02
0.01
0.00
20
30
40
50
Parameter
60
70
80
FIGURE 4-26 This figure shows that the normal distribution with x 5 50 and s 5 50 , closely conforms
to the Poisson distribution with m 5 50.
85
imaging as well. If an x-ray beam with a photon fluence f (photons/mm2) is

incident on a digital x-ray detector system with detection efficiency e and with
detector elements of area A (in mm2), the average number of x-rays recorded in
each detector element will be N, where f A e 5 N. Using the normal distribution
assumption for the Poisson distribution, and now using N to represent the mean
number of photons (instead of the symbol x used previously) in the image, the
noise per pixel is given by

N
[4-17]
The above equation is the stipulation that allows the normal distribution to
approximate the Poisson distribution. In other words, we are constraining the twoparameter normal distribution so that it now behaves like a one-parameter distribution. This allows us to adjust the technique parameters (N) in x-ray imaging a priori,
to control the noise levels perceived in the image. The relative noise on an image is
equal to the coefficient of variation (COV):
Relative noise COV

[4-18]
N
So, we can adjust N in an x-ray image, as it is proportional to various settings
on the x-ray system discussed later in this text. In nuclear imaging, N is determined
by the amount of isotope injected and the scan time, amongst other factors. In
general (at the same energy), N is also linearly proportional to the radiation dose
to the patient. By adjusting N, we are also adjusting s (Equation 4-17), which in
turn means that we are adjusting the relative noise in the image (Equation 4-18).
Therefore, we can predict the relative noise in the image and use this knowledge to
adjust the value of N (and hence s and COV) for the clinical imaging task, before
we acquire that image.
A clinical example will help illustrate the power of Equation 4-18. In fluoroscopy,
we only need to see a catheter advance in a vessel for part of the interventional vascular procedure, so we know that we can use a relatively small value of N (per pixel per
image) for this task, acknowledging that the fluoroscopic images will be noisy, but
experience has shown that this is adequate for providing the radiologist the image
quality necessary to advance the catheter. This saves dose to the patient for what can
ultimately be a time-consuming, high-dose procedure. By comparison, mammography requires very low relative noise levels in order to determine, from a statistical
standpoint, that there are small microcalcifications on the image. This means that we
need to use a relatively high value of N per pixel for breast imaging. The fluoroscopic
and mammography systems are adjusted to deliver a certain N per pixel in the resulting images, but it is the Poisson Distribution, and its handier normal distribution
approximation, that allow us to predict the noise in the image from the mean photon
fluence to the detector. The bottom line is that because the noise level in an image
(s/N) is determined by setting N, this allows the compromise between radiation dose
to the patient and noise in the image to be determined a priori.
There are very important dose ramifications when working in an environment
where the Poisson distribution is in play: Doubling the radiation dose to the patient
implies adjusting the photon fluence from N to 2N, but since = N , this will
reduce the noise by a factor of 2 51.41. Hence, doubling the dose results in a
41% reduction in noise. To reduce the noise by half, a fourfold increase in radiation
dose to the patient is needed. Table 4-1 illustrates how noise and SNR change as a
function of N.
86
TABLE 4-1 EXAMPLES OF NOISE VERSUS PHOTONS

NOISE ()
RELATIVE NOISE
SNR
PHOTONS/PIXEL (N)
( =
N)
(/N) (%)
(N/)
10
3.2
32
3.2
100
10
10
10
1,000
31.6
3.2
32
10,000
100
1.0
100
100,000
316.2
0.3
316
SNR, signal-to-noise ratio.
4.6
Noise Texture: The Noise Power Spectrum

The measurement that is characterized by the variance, s2 is a simple metric which
can quantify the noise in an image using Equation 4-14, but this metric does not
quantify the noise texture. In Figure 4-27, the two CT images of a test object have
the same variance in the background, but there is a perceptible difference in appearance of the way the noise looks. Although the noise variance is the same, the frequency dependence of the noise is different. The frequency dependence of the
noise variance is characterized by the noise power spectrum, NPS(f), where for a
2D image I(x,y)
NPS (fx , f y ) =
I ( x , y ) I e
x y
2 i ( xfx + yfy )
dx dy
[4-19]
where fx is the frequency corresponding to the x-dimension and fy is the frequency corresponding to the y-dimension, and I is the mean of image I(x,y). Just as
the MTF(f) gives a richer, frequency-dependent measure of how an imaging system
operates on an input signal, the NPS(f) yields an informative, frequency-dependent
measure of how an imaging system operates on the noise input into the system. The
NPS is essentially a frequency-dependent breakdown of the variance, and indeed the
integral of the NPS over all frequencies equals the variance s2.
FIGURE 4-27 Two CT images of a test object are shown, and the standard deviation in the highlighted
boxes is identical. However, the noise texturethe way the noise looksis different. These differences in
noise texture are characterized using the frequency dependent noise power spectrum, NPS(f).
87
25
uncorrelated, white noise
20
NPS(f)
15
correlated noise
10
0
0
Spatial Frequency (cy / mm)

FIGURE 4-28 The noise power spectrum, NPS(f), is shown for uncorrelated noise and for correlated noise.
2 =
fx
fy
NPS ( f x , f y ) df x df y
[4-20]
If the noise in each pixel of a 2D image is not dependent upon the noise values
in any of its surrounding pixels, then there will be no noise correlation and the NPS(f)
will essentially be a flat, horizontal line (Fig. 4-28). This type of uncorrelated noise
is called white noise. Real imaging systems have some blur phenomenon that results
in the finite width of the PSF(x) or LSF(x). This blurring means that noise from
detector elements can leak into the adjacent detector elements, leading to noise correlation between adjacent pixels in the image. There are many types and causes of
noise correlation (including anticorrelation, where positive noise in one pixel will
tend to induce a negative noise value in adjacent pixels), including reconstruction
algorithms in tomography, but in general, the result of noise correlation is that the
NPS is no longer whiteand the shape of the NPS(f) for a given imaging system
then is a technical description of this broader sense of noise texture (see Fig. 4-28).
The noise power spectrum is an analytical tool that is used by imaging scientists
but is not generally used in the clinical radiology setting. It is an important metric
when considering the design of new imaging systems, and in the comprehensive evaluation of research imaging systems. Nevertheless, some familiarity with the concept of
the NPS is useful to the clinically focused reader, because reconstructed images such
as in CT have a wide array of noise textures that are plainly visible on clinical images,
and these textures depend on the reconstruction methodology and kernels used.
We will return to concepts of the NPS later in this chapter.
4.7
Contrast
Subject Contrast
Subject contrast is the fundamental contrast that arises in the signal, after it has
interacted with the patient but before it has been detected. The example of projection
radiography is illustrated in Figure 4-29. In the case of x-ray projection imaging, an
88
FIGURE 4-29 An x-ray beam is incident on a simple patient, and the incident beam of x-ray photons is modulated
by attenuation processes in the patient,
resulting in a heterogeneous pattern of
x-rays emerging from the patient, which
then are incident upon the x-ray detector.
Subject contrast is defined as the differences in the x-ray beam fluence across
the beam, emmerging from the patient
which is incident upon the detector.
Homogeneous Incident
X-ray Beam
Patient
Detector
A
B
Profile of X-ray beam

emerging from patient
prior to reaching
detector
approximately homogeneous x-ray beam is incident upon the patient, and the x-rays
then interact via various interaction mechanisms, resulting in the majority of x-rays
being attenuated, but a small fraction will emerge from the patient unattenuated.
Although this beam cannot be measured in reality until it reaches a detector, the
concept of subject contrast involves the fundamental interaction between the x-ray
beam and the object. The subject contrast is defined as

Cs
(A B)

A
[4-21]
where the photon fluence levels of A and B are shown in Figure 4-29, and for Equation
4-21, A . B. This requirement means that contrast runs from 0 to 1 (i.e., 0% to 100%).
Subject contrast has intrinsic factors and extrinsic factorsthe intrinsic component of subject contrast relates to the actual anatomical or functional changes in
the patients tissues, which give rise to contrast. That is, the patient walks into the
imaging center with intrinsic, physical or physiological properties that give rise to
subject contrast. For a single pulmonary nodule in the lung, for example, the lesion
is of greater density than the surrounding lung tissue, and this allows it to be seen on
chest radiography or thoracic CT or thoracic MRI. The lesion may also exhibit higher
metabolism, and thus, when a sugar molecule labeled with radioisotope is injected
into the patient, more of that biomarker will accumulate in the lesion than in the surrounding lung due to its greater metabolism, and the radioisotope emission resulting
from this concentration difference results in subject contrast.
Extrinsic factors in subject contrast relate to how the image-acquisition protocol can be optimized to enhance subject contrast. Possible protocol enhancements
include changing the x-ray energy, using a different radiopharmaceutical, injecting
an iodine (CT) or gadolinium (MR) contrast agent, changing the delay between contrast injection and imaging, changing the pulse sequence in MR, or changing the
angle of the ultrasound probe with respect to the vessel in Doppler imaging. These
are just a few examples of ways in which subject contrast can be optimized; many
more parameters are available for optimization.
Detector Contrast
The incident beam of energy from the imaging system will eventually reach the
detector(s), and again radiography is used as an example (Fig. 4-30). The same log
89
Signal Levels on Detector 2
Signal Levels on Detector 1
shoulder
toe
A
Input Signal
FIGURE 4-30 For a log input signal (x-axis) with two identical objects attenuating the beam at different
fluence levels in the beam (inputs A and B), the signal levels on Detector 1 (screen-film system) are dramatically influenced by the nonlinear characteristic curve of the detector. By comparison, the linear detector system
(detector 2) produces signals that are proportional to the input signal trends. Detector 1 amplifies the signal
contrast in the steep area of the curve, while a reduction in signal contrast is realized in the toe and shoulder of
detector 1sc haracteristic curve.
relative exposure differences are shown on the x-axis, and two different detector
responses are showna nonlinear response to x-rays (screen-film radiography) is
shown on the left vertical axis, and a linear response (typical of many digital x-ray
detectors) is shown on the right vertical axis. This figure illustrates that the detector
system can have a profound impact on how the subject contrast incident upon it is
converted to the final image.
Displayed Contrast
The detector contrast acts to modulate the subject contrast, but the acquisition procedure ultimately results in the capture of digital image data on a computers hard
drive. For modalities where subsequent image processing occurs, such as image processing in many digital radiography systems, or image reconstruction in CT, MRI,
SPECT, PET, etc., the raw image information is processed into an image that is finally
meant for physician viewing (for presentation images, in the parlance of PACS).
The majority of medical images have bit depths ranging from 10, 12, and even
14 bits, which run from 1,024, 4,096, to 16,384 shades of gray, respectively. Modern
displays, however, are only capable of displaying 8-bit (256 shades of gray) to 10-bit
(1,024 shades of gray) images. The display computer needs to convert the higher
bit depth data encoded on the image to the spectrum of gray scale on the monitor,
and there is a look-up table (LUT) that is used to make this transformation. Various
LUTs are illustrated in Figure 4-31. The LUT transforms the gray scale of each pixel
of the image stored in memory on the computer, ultimately to be displayed as gray
scale on the monitor. The most commonly used LUT in radiological imaging is the
so-called window/level, which is depicted in Figure 4-31. This is generally a nonlinear LUT that causes saturation to black below L2W/2 and saturation to white above
90

Window
- white
level
- black
P1
P2
FIGURE 4-31 A digital image is represented as a matrix of gray scale values in the memory of a computer,
but the look-up table (LUT) describes how that gray scale is converted to actual densities on the display. Three
different window/level settings are shown on the left. The diagram shows the window width and level, and
the image data is saturated to black at P1 5 L 2 W/2, and is saturated white at P2 5 L 1 W/2. Chest radiographs are shown with different window/level settings.
L1 W/2. Because the monitor cannot depict the depth of gray scale from the entire
image, it is routine for the interpreting physician to routinely change the window/
level settings for a given image, so that the entire range of gray scale on the acquired
image can be evaluated. While freely changing the window and level setting is possible on most image workstations, several preset window and level settings are generally used to expedite viewing.
There is another approach for medical image display, which reduces the need for
the physician to view images using multiple LUT settings. Simple image processing
techniques can reduce the dynamic range of the native image for display down to a
point where all gray scale in the image can be viewed using one window and level
settingone LUT (Fig. 4-32). While there are many examples of this, the most obvious is that of digital mammographywhere the raw 14-bit image is collapsed down
to a displayable 10-bit image using image processing methods akin to blurred mask
subtraction.
3000
Gray Scale
2500
2000
1500
1000
500
0
Original Image
Relative Position Across Image
Processed Image
FIGURE 4-32 A standard chest x-ray is shown on the left, and the corresponding trace through it is shown
in the graph. Image processing methods such as blurred mask subtraction or other high pass filtering techniques can be used to compress the gray scale necessary for the display of the image (right image). The gray
scale versus position trace of the processed image has less dynamic range than in the original image.
91
Mean BG = 127.5332
Noise = 10.2949
Steps are 5, 10, 15, 20
FIGURE 4-33 The CNR is illustrated. The CNR is an area-independent measure of the contrast, relative to the
noise, in an image. The CNR is useful for optimizing image acquisition parameters for generic objects of variable
sizes and shapes. The CNRs in this figure (in the disks running from left to right) are 0.39, 1.03, 1.32, and 1.70.
4.8
Contrast-to-Noise Ratio
The contrast-to-noise ratio (CNR) is an object size-independent measure of the signal
level in the presence of noise. Take the example of a disk as the object (Fig. 4-33). The
contrast in this example is the difference between the average gray scale of a region
of interest (ROI) in the disk (x S ) and that in an ROI in the background (x BG) , and the
noise can be calculated from the background ROI as well. Thus, the CNR is given by
CNR =
(x S x bg )

bg
[4-22]
The CNR is a good metric for describing the signal amplitude relative to the
ambient noise in an image, and this is particularly useful for simple objects. Because
the CNR is computed using the difference in mean values between the signal region
and the background, this metric is most applicable when test objects that generate a
homogeneous signal level are usedthat is, where the mean gray scale in the signal
ROI is representative of the entire object. Example uses of the CNR metric include
optimizing the kV of an imaging study to maximize bone contrast at a fixed dose
level, computing the dose necessary to achieve a given CNR for a given object, or
computing the minimum concentration of contrast agent that could be seen on given
test phantom with fixed dose.
4.9
Signal-to-Noise Ratio
The SNR is a metric similar to the CNR, except that the size and shape of the object
is explicitly included in the computation. In addition, the SNR does not require the
test object that generates the signal to be homogeneous; however, the background
92
FIGURE 4-34 Signal regions with a gaussian-based intensities are shown. The SNR has a fundamental
relationship with detectability, and the Rose Criterion states that if SNR . 5, the signal region will be detectable in most situations.
does need to be homogeneous in principlea series of gaussian blobs is used to

illustrate this point (Fig. 4-34). The numerator in the SNR is the signal integrated
over the entire dimensions of the object of interest. The signal amplitude of each
pixel is the amount in which this patch of the image is elevated relative to the
mean background signalthus, for a mean background of x BG , the signal at each
pixel i in the image is xi 2 x BG . The denominator is the standard deviation in the
homogeneous background region of the image (sBG), and thus the SNR represents
the integrated signal over an ROI, which encapsulates the object of interest, divided
by the noise

SNR =
(x x
i
BG
BG
[4-23]
The SNR can be computed by summing the difference values over a rectangular
ROI, which surrounds the signal region. Notice that this definition of the SNR allows
the computation to be performed even when the signal region is not homogeneous
in gray scale (see Fig. 4-34). The SNR computed in Equation 4-23 does require that
the measurement of x BG be accurate, and thus it should be computed over as large
a region as possible.
The SNR is one of the most meaningful metrics that describes the conspicuity
of an objecthow well it will be seen by the typical observer. Indeed, Albert Rose
recognized this and was able to demonstrate that if SNR 5, then an object will
almost always be recognized (detected), but that detection performance continuously degrades as SNR approaches zero. This is called the Rose Criterion.
4.10 Contrast-Detail Diagrams

In this chapter, we have discussed spatial resolution and contrast resolution as if they
were separate entities, but of course both of these quantities matter on a given image.
In particular, it does not matter if the imaging receptor has excellent spatial resolution if the noise in the detector is high (i.e., the contrast resolution is low) and the
Detail
93
Contrast
Contrast
Contrast
Detail
Detail
FIGURE 4-35 Contrast-detail diagrams are illustrated. A. The (noiseless) CD phantom is illustrated, where
disks are smaller to the left and have less contrast toward the bottom. B. Some resolution loss and added noise
is present, and the smallest and lowest in contrast disk can no longer be seen with confidence. The yellow line
is the line of demarcation between disks that can be seen (upper right) and those that cannot be seen (lower
left). C. More noise is added, and more of the subtle disks (including the entire, low-contrast bottom row)
cannot be reliably seen.
statistical integrity of the image is not sufficient to detect a small object. The contrast
detail diagram, or CD diagram, is a conceptual, visual method for combining the
concepts of spatial resolution and contrast resolution.
A standard CD phantom is illustrated in Figure 4-35A. Here, the disk diameter
decreases toward the left, so the greatest detail (smallest disks) is along the left column of the CD diagram. The contrast of each disk decreases from top to bottom, and
thus the bottom row of the CD diagram has the lowest contrastand the disks in
the bottom row will therefore be most difficult to see in a noisy image. Figure 4-35B
illustrates the CD diagram, but with some resolution loss and with increased noise.
It is important to remember that the most difficult disk to see is in the lower left
(smallest with lowest contrast), and the easiest disk to see is the upper right (largest
with highest contrast) on this CD diagram. The line (see Fig. 4-35B) on a CD diagram
is the line of demarcation, separating the disks that you can see from the ones that
you cannot see. To the left and below the line are disks that cannot be seen well, and
above and to the right of the line are disks that can be seen. Figure 4-35C shows the
CD diagram with even more noise, and in this image the line of demarcation (the
CD curve) has changed because the increased noise level has reduced our ability to
see the disks with the most subtle contrast levels.
Figure 4-35 illustrates CD diagrams with the actual test images superimposed
behind the graph; however, Figure 4-36 illustrates an actual CD diagram. The two
curves in Figure 4-36A correspond to imaging systems A and B, but how can the differences in these two imaging systems be interpreted? Let curve A in Figure 4-36A
be a standard radiographic procedure and curve B represents a low-dose radiograph.
The dose does not change the detail in the image, and thus the two curves come
together in the upper left (where the contrast is the highest) to describe the same
detail level. However, because curve A corresponds to an image that was acquired
with higher radiation dosethe noise in the image is less, and the CD diagram
shows that curve A has better contrast resolutionbecause (lower contrast) disks
lower down on the CD diagram can be seen (relative to curve B).
Figure 4-36B shows another example of a CD diagram. The two curves A and B
on this CD diagram describe two different image processing procedures for the same
image. Let curve B be the original acquired data of the CD phantom on some imaging modality, and curve A has been smoothed using image processing techniques
(see Fig. 4-11). The smoothing process blurs edges (reduces detail) but reduces image
94
Contrast
Contrast
B
B
Detail
Detail
FIGURE 4-36 A. CD curves are shown. Both curves demonstrate the same detail (in the limit of highcontrast objects), but curve A extends to lower contrast. One explanation is that curve A corresponds to a
higher dose version of curve B. B. Curve A has less detail but more contrast resolution than Curve B, and
one explanation would be that Curve A corresponds to an image that was smoothed, relative to the image
corresponding to curve B.
noise as well (increases contrast resolution). Thus, relative to curve B, curveA has
less detail (does not go as far left as curve B) due to the blurring procedure, but has
better contrast resolution (curve A goes lower on the CD diagram) because blurring
smoothes noise.
The use of CD diagrams unites the concepts of spatial resolution (i.e., detail) and
contrast resolution (i.e., SNR) on the same graph. It is a subjective visual test, and so
it is excellent in conveying the relationships visually but it is not quantitative. The
detective quantum efficiency (DQE), discussed in the next section, also combines the
concepts of spatial resolution and contrast resolution; however, the methods of DQE
analysis are less visual and more quantitative.
4.11 Detective Quantum Efficiency

The DQE of an imaging system is a characterization of an x-ray imaging system, used
by imaging scientists, which describes the overall frequency-dependent SNR performance of the system. In conceptual terms, the DQE can be defined as the ratio of the
SNR2 output from the system to the SNR2 of the signal input into the system

DQE(f)
SNR 2OUT

SNR 2IN
[4-24]
The SNRIN to an x-ray imaging system is simply SNRIN 5 N = N N = N (Equation

4-17), and thus SNR2IN 5 N, the mean photon fluence incident upon the imaging
system.
The SNR2OUT is a function of the MTF and the NPS. The MTF(f) describes how
well an imaging system processes signal, and the NPS(f) describes how well an imaging
system processes noise in the image.
Combining these concepts, the numerator of Equation 4-19 is given by

SNR 2OUT
[MTF(f)]2

NPS(f)
[4-25]
95
The NPS(f) is the noise variance, so it is already squared. Equation 4-25 also
describes the Noise Equivalent Quanta (NEQ).
With the above definitions for SNR2IN and SNR2OUT, the DQE(f) is given by
DQE(f)
k [MTF(f)]2

N NPS(f)
[4-26]
where k is a constant that converts units, and its value is usually defined as the square
of the mean pixel value in the image(s) used to measure NPS(f). Its role is to convert
the gray scale units of the imaging system to generic relative noise so as to be comparable with N (the Poisson noise term). The NPS(f) is determined from an image (or
images) acquired at a mean photon fluence of N photons/mm2. Thus, the units of
the DQE(f) are mm22. The DQE(f) for a given detector system is shown as a function
of different exposure levels to the detector in Figure 4-37. The DQE(f) has become
the standard by which the performance of x-ray imaging systems is measured in the
research environment. The DQE(f) is an excellent description of the dose efficiency of
an x-ray detector systemthat is how well the imaging system converts SNR2 incident on the detector into SNR2 in the image. The DQE(f), is a frequency-dependent
description, and at zero frequency (f 5 0) in the absence of appreciable electronic
noise, the DQE(0) essentially converges to the detectors quantum detection efficiency,
the QDE. The QDE reflects the efficiency of x-ray detection, neglecting other elements
in the imaging chain that inject or amplify noise. It can also be estimated using
QDE =
f(E ) 1 e
E
f (E )
( E )T
dE
dE
[4-27]
where f(E) is the x-ray spectrum, m(E) represents the energy dependent linear attenuation coefficient of the x-ray detector material, and T is the thickness of the detector
material. For a monoenergetic x-ray beam, Equation 4-27 collapses to the term in
the square brackets.
1.0
6.79 Gy / frame
0.9
4.85 Gy / frame
0.8
2.91 Gy / frame
0.7
0.97 Gy / frame
DQE
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.0
0.2
0.4
0.6
0.8
1.0
Spatial frequency (cycles / mm)
1.2
1.4
FIGURE 4-37 The DQE(f), is shown for a flat panel detector used on a cone beam CT system is shown. Four
curves were generated at different exposure levels per image, as indicated. The curve at the lowest exposure
level (0.97 mGy/frame) shows reduced DQE(f), likely due to the presence of added electronic noise. The electronic noise is present in the other curves as well, but the higher exposure levels dominate the signal intensity,
and thus the electronic noise has less of an influence in degrading performance.
96
4.12 Receiver Operating Characteristic Curves

Image quality can be quantified using a number of metrics that have been discussed
above. Ultimately, however, the quality of an image relates to how well it conveys
diagnostic information to the interpreting physician. This can be tested using the
concept of the receiver operating characteristic (ROC) curve. ROC performance
includes not only the quality of the image data produced by an imaging system for
a specific task, but in some cases the skill of the interpreting physician in assessing
this data is also included.
The starting point of ROC analysis is the 2 3 2 decision matrix (sometimes called
the truth table), shown in Figure 4-38. The patient either has the suspected disease
(actually abnormal) or not (actually normal), and this gold standard is considered
the truth. The diagnostician has to make a binary decision about whether he or she
thinks the patient has the suspected disease or notbased upon one or more diagnostic tests (including imaging examinations), he or she calls the patient either normal
(does not have the disease) or abnormal (has the disease). The 2 3 2 decision matrix
defines the terms true positive (TP), true negative (TN), false positive (FP), and false negative (FN) (see Fig. 4-38). Most of the work in performing patient-based ROC studies
is the independent confirmation of the truth, and this may require biopsy confirmation, long-term patient follow-up, or other methods to ascertain the true diagnosis.
In most radiology applications, the decision criterion is not just one feature or
number, but is rather an overall impression derived from a number of factors, sometimes referred to as gestaltessentially, the gut feeling of the experienced radiologist.
Figure 4-39A illustrates two distributions of cases, those that are normal (left curve)
and those that are abnormal (right curve). The generic descriptors normal and abnormal are used here, but these can be more specific for a given diagnostic task. For
example, for mammography, the abnormal patients may have breast cancer and the
normal patients do not. Overlap occurs between normal and abnormal findings in
imaging, and thus these curves have overlap. Even though there is some ambiguity
in the decision due to the overlap of the two curves, the radiologist is often responsible for diagnosing each case one way or another (e.g., to make the decision to
biopsy or not). Hence, the patient is either determined to be normal or abnormala
binary decision. In this case the diagnostician sets his or her own decision threshold
(Fig. 4-39A). Patients to the right of the decision threshold are considered abnormal,
and patients to the left of the threshold are considered normal. When the threshold
is set, the values of TP, TN, FP, and FN can be determined (see Fig. 4-39).
FIGURE 4-38 The 2 3 2 decision matrix is illustrated.
actually
abnormal
normal
TN
FN
abnormal
Observer
Response
Truth
actually
normal
FP
TP
97

1.0
Number of Cases
60
50
normal
40
TN
FP
30
20
0.8
0.7
0.6
AUC
0.5
0.4
0.3
0.2
TP
0.1
abnormal
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
FN
10
0
True Positive Fraction
0.9
10 20 30 40 50 60 70 80 90 100
Decision Parameter
False Positive Fraction
FIGURE 4-39 A. The underlying data that explain ROC analysis is shownthis data is not always available
for the actual curve generation. Histograms of normal and abnormal populations of patients are illustrated, but
the observer needs to set a binary decision threshold (vertical line, T) such that cases to the right are called
abnormal, and cases to the left of the threshold are diagnosed as normal. The two populations and the single
threshold value define the TP, FN, TP, and FP. These values are used to compute sensitivity and specificity, as
defined in the text, which are used to define one point on the ROC curve (Fig. 4-39B). To compute the entire
ROC curve, the threshold needs to be swept across the x-axis, which will lead to the computation of many pairs
of points needed to draw the entire ROC curve. B. An ROC curve is defined as the TPF (sensitivity) as a function
of FPF (1 specificity), and the ROC curve corresponding to the distributions shown in Figure 4-39A is shown.
The black dot is the point on the curve defined by the current position of the threshold (T) on Figure 4-39A.
The dotted line shows an ROC curve for an observer who merely guesses at the diagnosisit represents the
minimum performance. The area under the curve (AUC) is often used as a shorthand metric for the overall
performance shown on an ROC curve. AUC values run from 0.5 (dotted line, pure guessing) to 1.0a perfect
ROC curve runs along the left and top axes, bounding the entire area of the unit square.
Using the values of TP, TN, FP, and FN, the true-positive fraction (TPF), also
known as the sensitivity, can be calculated as

TPF sensitivity
TP

TP FN
[4-28]
The true-negative fraction, the specificity, can also be determined as follows:

TNF specificity
TN

TN FP
[4-29]
The false-positive fraction (FPF) can be determined as well:

FPF
FP ,

TN FP
[4-30]
It can be seen that the FPF 5 1 2 TNF 5 1 2 specificity.

An ROC curve (see Fig. 4-39B) is a plot of the TPF as a function of the FPF, which
is also the sensitivity as a function of (1-specificity). It can be seen in Figure 4-39A
that the selection of one threshold value results in defining specific values of TP,
FP, TN, FN such that one value of sensitivity (Equation 4-28) and one value of
specificity (Equation 4-29) can be calculatedthis contributes to one point on the
ROC curve (the black dot in Figure 4-39B). In order to compute the entire ROC curve
(Fig. 4-39B), which is defined by many pairs of (FPF, TPF) values, the threshold value
(T) in Figure 4-39A has to be swept over the entire width of the two distributions.
98
For some extrinsic ROC computations where (often, computer-generated) data

similar to that shown in Figure 4-39A are available, the threshold T can be stepped
across the data sets and a smooth ROC curve can be computed (as was done to make
these plots). For intrinsic ROC studies involving human observers, the data similar
to that shown in Figure 4-39A are not available. Instead, the human observer is
asked to use five or more different confidence levels for the decision threshold, and
classically these are given the labels: definitely not there, probably not there, uncertain,
probably there, and definitely there. Using the data derived from this test, five different
2 3 2 truth tables can be established, resulting in five different pairs of sensitivityspecificity values. These five points can be used to estimate the full ROC curve, and
for scientific analysis, a mathematical ROC model is typically used to fit the data.
The ROC curve is plotted on a unit square (area 5 1.0), and it turns out that pure
guessing results in a diagonal line (Fig. 4-39B, dashed line). The area under the ROC
curve is often referred to as simply AUC, and this value is a number that ranges from
0.50 (the diagonal line of pure guessing) to 1.00. The AUC value is often referred to
as a shorthand metric of observer performance. The AUC of the ROC curve shown
in Figure 4-39B is 0.77.
The intrinsic ROC curve is a description of the performance of a given observer
at a given task (e.g., Dr. Smith in mammography, Dr. Jones in head MRI). Although
the ROC curve describes their theoretical performance as if they were able to dramatically vary their threshold value, in reality Dr. Smith has a typical operating point at a
diagnostic task such as mammography, and this operating point resides somewhere
along the ROC curve. Take the ROC curve shown in Figure 4-40, which has an
AUC of about 0.93, a realistic value for screening mammography. It is known that
the call back rate in screening mammography is approximately 6% (depends on the
radiologist), and this corresponds very nearly to a specificity of 94% (100 2 6). The
operating point at this specificity is shown in Figure 4-40, and it can be seen that the
corresponding sensitivity is about 70%. The radiologists can move their operating
point along the ROC curve, but for an experienced radiologist, the curve is pretty
much fixed for a given modality and patient population. To increase their sensitivity,
the specificity would have to decrease, increasing the call back rate and probably
increasing the negative biopsy rate as well. These observations suggest that there is a
trade off that diagnosticians make in setting the operating point on their ROC curve.
The widespread use of the parameters specificity and sensitivity in the medical
literature is partly because these parameters are independent of disease incidence.
Notice in Figure 4-39A, the abnormal population is smaller than the normal population (the area of the abnormal curve is smaller). The ROC curve computation (using
sensitivity and specificity metrics) remains the same, regardless of the relative areas
of the normal and abnormal populations. There are a number of other parameters
related to observer performance that are not independent of incidence.
Accuracy is defined as

Accuracy
TP TN

TP TN FP FN
[4-31]
It is tempting to use accuracy as a measure of diagnostic performance, but accuracy is highly influenced by disease incidence. For example, due to the low incidence
of breast cancer in a screening population, a breast imager could invariably improve
accuracy by just calling all cases negative, not even looking at the mammograms.
Because the incidence of breast cancer in the screening population is low (3/1,000),
with all cases called negative (normal), TN would be very large (997 per 1,000),
FP and TP would be zero (since all cases are called negative, there would be no
99
1.0
0.9
True Positive Fraction
0.8
Sensitivity (~70 %)
0.7
0.6
0.5
Specificity (94 %)
0.4
0.3
0.2
AUC = 0.93
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
False Positive Fraction
FIGURE 4-40 An ROC curve is shown, and this curve is plausible for breast cancer screening using mammography. The ROC curve describes the theoretical performance of a skilled observer, but this observer will
generally operate at a given point on the curve, called the operating point (black dot). This figure demonstrates the trade-offs of decision performancecompared to the operating point shown, the radiologist could
increase sensitivity but generally this will come at a reduction in specificitythe operating point will move
upward and to the right along the curve. More cancers will be detected (sensitivity will be increased), but far
more women will be called back and asked to undergo additional imaging procedures and potentially biopsies
(reduced specificity).
ositives), and accuracy would be approximately 99.7%. Although the specificity

p
would be 100%, the sensitivity would be 0%obviously a useless test in such a case.
For these reasons, diagnostic accuracy defined in Equation 4-31 is rarely used as a
metric for the performance of a diagnostic test.
The positive predictive value (PPV) refers to the probability that the patient
is actually abnormal (TP), when the diagnostician says the patient is abnormal
(TP 1 FP).

Positive predictive value
TP

TP FP
[4-32]
Conversely, the negative predictive value (NPV) refers to the probability that the
patient is actually normal (TN), when the diagnostician says the patient is normal
(TN 1 FN).

Negative predictive value
TN

TN FN
[4-33]
The PPV and NPV of a diagnostic test (such as an imaging examination with radiologist interpretation) are useful metrics for referring physicians as they weigh the
information from a number of diagnostic tests (some which may be contradictory)
for a given patient, in the process of determining their diagnosis.
SUGGESTED READING
Barrett HH, Swindell W. Radiological imaging: the theory of image formation, detection, and processing, vols. 1
and 2. New York, NY: Academic Press, 1981.
100

Bracewell RN. The Fourier transform and its applications. New York, NY: McGraw-Hill, 1978.
Boone JM. Determination of the presampled MTF in computed tomography. Med Phys 2001;28:356360.
Dainty JC, Shaw R. Image science. New York, NY: Academic Press, 1974.
Fugita H, Tsai D-Y, Itoh T, et al. A simple method for determing the modulation transfer function in digital
radiography, IEEE Trans Med Imaging 1992;MI-11:3439.
Hasegawa BH. The physics of medical x-ray imaging, 2nd ed. Madison, WI: Medical Physics, 1991.
Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978;8:283298.
Rose A. Vision: human and electronic. New York, NY: Plenum Press, 1973.
Chapter
Medical Imaging Informatics

Medical imaging informatics is a multidisciplinary field of science and engineering
that addresses the gathering, transfer, storage, processing, display, perception, and
use of information in medicine. It overlaps many other disciplines such as electrical
engineering, computer and information sciences, medical physics, and perceptual
physiology and psychology.
This chapter begins by discussing number systems, the analog and digital representation of information, and the conversion of information between analog and digital representations. It next discusses digital radiological images and briefly describes
digital computers and digital information storage technology. After this is a discussion of the display of digital radiological images. This is followed by a description of
computer networks, which permit the rapid transfer of information over short and
long distances. The next section addresses picture archiving and communications
systems (PACS), which store and supply digital medical images for display. The last
sections review image processing methods and measures for the security of medical
information stored in digital form or transferred using computer networks.
5.1 Analog and Digital Representation of Data
Number Systems
Our culture uses a number system based on ten, probably because humans have
five fingers on each hand and number systems having evolved from the simple act
of counting on the fingers. Computers use the binary system for the storage and
manipulation of numbers.
Decimal Form (Base 10)

In the decimal form, the ten digits 0 through 9 are used to represent numbers. To
represent numbers greater than 9, several digits are placed in a row. The value of each
digit in a number depends on its position in the row; the value of a digit in any position is ten times the value it would represent if it were shifted one place to the right.
For example, the decimal number 3,506 actually represents
(3 103 ) (5 102 ) (0 101 ) (6 100 )
where 101 10 and 100 1 . The leftmost digit in a number is called the most significant digit and the rightmost digit is called the least significant digit.
Binary Form (Base 2)

In binary form, the two digits 0 and 1 are used to represent numbers. Each digit, by
itself, has the same meaning that it has in the decimal form. To represent numbers
greater than 1, several digits are placed in a row. The value of each digit in a number
101
102
TABLE 5-1NUMBERS IN DECIMAL AND BINARY
FORMS
DECIMAL
BINARY
DECIMAL
BINARY
1000
1001
10
10
1010
11
11
1011
100
12
1100
101
13
1101
110
14
1110
111
15
1111
16
10000
depends on its position in the row; the value of a digit in any position is two times
the value it would represent if it were shifted one place to the right. For example, the
binary number 1101 represents
(1 23 ) (1 22 ) (0 21 ) (1 20 )
3
where 2 8 , 22 4 , 21 2 , and 20 1 .
To count in binary form, 1 is added to the least significant digit of a number. If
the least significant digit is 1, it is replaced by 0 and 1 is added to the next more
significant digit. If several contiguous digits on the right are 1, each is replaced by
0 and 1 is added to the least significant digit that was not 1. Counting in the binary
system is illustrated in Table 5-1.
Conversions Between Decimal and Binary Forms

To convert a number from binary to decimal form, the binary number is expressed
as a series of powers of two and the terms in the series are added. For example, to
convert the binary number 101011 to decimal form
101011 (binary) 5 (1 3 25) 1 (0 3 24) 1 (1 3 23) 1 (0 3 22) 1 (1 3 21) 1 (1 3 20)
5 (1 3 32) 1 (1 3 8) 1 (1 3 2) 1 (1 3 1) 5 43 (decimal).
To convert a number from decimal into binary representation, it is repeatedly divided
by two. Each division determines one digit of the binary representation, starting
with the least significant. If there is no remainder from the division, the digit is 0; if
the remainder is 1, the digit is 1. The conversion of 42 (decimal) into binary form is
illustrated in Table 5-2.
Considerations Regarding Number Systems

Whenever it is not clear which form is being used, the form is written in parentheses
after the number or denoted by a subscript of 2 for binary form and 10 for decimal
form. For example, the number five can be written as 101 (binary), 1012, 5 (decimal), or 510. If the form is not specified, it can usually be assumed that a number is
in decimal form. It is important not to confuse the binary representation of a number with its more familiar decimal representation. For example, 10 (binary) and 10
(decimal) represent different numbers, although they look alike. On the other hand,
Chapter 5 Medical Imaging Informatics
103
TABLE 5-2CONVERSION OF 42 (DECIMAL) INTO
BINARY FORM
DIVISION
RESULT
REMAINDER
42/2
21
21/2
10
10/2
5/2
2/2
1/2
Least significant digit
Most significant digit
Note: The decimal value is repeatedly divided by 2, with the remainder recorded
after each division. The binary equivalent of 42 (decimal) is therefore 101010.
1010 (binary) and 10 (decimal) represent the same number. The only numbers that
appear the same in both systems are 0 and 1.
Analog and Digital Representations

The detectors of medical imaging devices inherently produce analog data. Image display devices, discussed later in this chapter, require image information in analog form
or must convert it to analog form. For many years, information in medical imaging
systems was in analog form from signal detection to image display. For example, during these years, information in fluoroscopes, nuclear medicine scintillation cameras,
and ultrasound imaging systems was entirely in analog form. Later, at some stages
in these devices, the information was converted into digital form for processing and
perhaps storage. Other imaging systems, such as x-ray computed tomography (CT)
systems, from inception converted analog information from the detectors into digital form for processing. Today, in nearly all radiological imaging systems, analog
information from the detectors is converted to digital form; processed, stored, and
transferred in digital form; and converted to analog form only in the last stages of the
image display systems.
Analog Representation of Data

In analog form, information is represented by a physical quantity whose allowed
values are continuous, that is, the quantity in theory can have an infinite number
of possible values between a lower and an upper limit. Most analog signals are also
continuous in time; such a signal has a value at every point in time. In electronic
circuits, numerical data can be represented in analog form by a voltage or voltage
pulse whose amplitude is proportional to the number being represented, as shown
in Figure 5-1A. An example of analog representation is a voltage pulse produced
by a photomultiplier tube attached to a scintillation detector. The amplitude (peak
voltage) of the pulse is proportional to the amount of energy deposited in the
detector by an x- or gamma ray. Another example is the signal from the video
camera attached to the image intensifier tube of a fluoroscopy system; the voltage
at each point in time is proportional to the intensity of the x-rays incident on a
portion of the input phosphor of the image intensifier tube (Fig. 5-1B). Numerical
data can also be represented in analog form in electronic circuits by the frequency
or phase of an alternating sinusoidal voltage, but these will not be discussed in
this chapter.
104

(0 decimal)
(5 decimal) (10 decimal)
0000
0101
1010
(10 volts)
10
V
(5 volts)
5
(0 volts)
0
0
10
20
30
Time (ms)
A
(80 decimal)
01010000
(200 decimal)
(40 decimal)
11001000
00101000
(200 mV)
200
mV 100
(80 mV)
(40 mV)
0
0
20
40
60
Time (ms)
B
FIGURE 5-1 Analog and digital representation of numerical data. A. Three analog voltage pulses, similar
to those produced by a photomultiplier tube attached to a scintillator, are illustrated. The height of each
pulse represents a number, for instance, energy. Data are sampled at 10, 20, and 30 ms in time. Above the
graph, these same amplitudes are represented in digital form with a sequence of four binary digits as on1,
off 0 binary states. Digital sampling occurs over a much shorter time scale than depicted on the x-axis.
B. A continuously varying analog signal, such as a single video trace from the video camera in a fluoroscopy
system, is shown. This analog signal varies in proportion to x-ray induced light intensities falling upon a photoconductive TV target, and produces an output voltage, in millivolts, as a function of time. The amplitude
of the signal is sampled and converted to a corresponding digital value; shown are digital samples at 20, 40
and 60 ms. The values are represented in digital form as sequences of 8 binary digits, with binary values that
correspond to the signal amplitude in mV.
Digital Representation of Information

Information in digital form is represented by a physical quantity that is allowed to
have one of only a limited number of discrete (separated) values at any moment in
time. In digital form, a number or other unit of information such as a letter of the
alphabet is represented by a sequence of digits, each of which is limited to a defined
number of values. Most digital signals are binary, that is, they consist of digits, each
of which is limited to only two allowed values, but digital signals that consist of digits
with more than two allowed values are possible and used in some situations.
The portion of a binary signal that is limited to one of two values is referred
to as a binary digit (bit). In an electronic circuit, a bit might be a voltage of either
0 or 5 V, maintained for a defined length of time. (In a practical digital circuit,
a range of voltages about each of these two voltages is permitted, but the two
ranges are separated by a wide gap.) In microprocessors, computer memory, and
mass storage devices, a bit is a physical entity. A bit in a microprocessor may
be an electronic switch that has two stable states. In most solid-state computer
memory, each bit is a capacitor; charged above a specific voltage is one state and
not charged is the other. In magnetic storage media, such as magnetic disks and
magnetic tape, a bit is a small portion of the disk or tape that may be magnetized
105
TABLE 5-3UNITS TO DESCRIBE COMPUTER MEMORY CAPACITY AND
INFORMATION STORAGE CAPACITY

Computer memory capacity
1 kilobyte (kB) 210 bytes 1,024 bytes < a thousand bytes
1 megabyte (MB) 220 bytes 1,024 kilobytes 1,048,576 bytes < a million bytes
1 gigabyte (GB) 230 bytes 1,024 megabytes 1,073,741,824 bytes < a billion bytes
Digital storage device or media capacity
1 kilobyte (kB) 103 bytes 1,000 bytes a thousand bytes
1 megabyte (MB) 106 bytes 1,000 kilobytes 1,000,000 bytes a million bytes
1 gigabyte (GB) 109 bytes 1,000 megabytes 1,000,000,000 bytes a billion bytes
1 terabyte (TB) 1012 bytes 1,000 gigabytes 1,000,000,000,000 bytes a trillion bytes
1 petabyte (PB) 1015 bytes 1,000 terabytes 1,000,000,000,000,000 bytes
Note: Note that the prefixes kilo-, mega-, and giga- have slightly different meanings when used to describe
computer memory capacity than in standard scientific usage, whereas they have the standard scientific
meanings when used to describe digital storage device or media capacity. To avoid confusion, some standards
organizations have advocated the units kibibyte (kiB, 1 kiB 210 bytes), mibibyte (MiB, 1 MiB 220 bytes),
and gibibyte (GiB, 1 GiB 230 bytes) for describing memory capacity.
in a specific direction. Because a bit is limited to one of two values, it can only
represent two numbers.
To represent more than two values, several bits must be used. For example, a
series of bits can be used to represent a number or another unit of information. The
binary number system allows a group of several bits to represent a number.
Bits, Bytes, and Words
As mentioned above, most digital signals and digital memory and storage consist
of many elements called bits (for binary digits), each of which can be in one of two
states. Bits are grouped into bytes, each consisting of eight bits. The capacity of a
computer memory, a storage device, or a unit of storage media is usually described
in kilobytes, megabytes, gigabytes, or terabytes (Table 5-3). (As noted in that table,
the prefixes kilo-, mega-, and giga-, when used to describe computer memory capacity, commonly have slightly different meanings than standard usage, whereas these
prefixes have their usual meanings when used to describe digital information storage
devices and media.) Bits are also grouped into words. The number of bits in a word
depends on the computer system; 16-, 32-, and 64-bit words are common.
Digital Representation of Different Types of Information
General-purpose computers must be able to store and process several types of information in digital form. For example, if a computer is to execute a word processing
program or a program that stores and retrieves patient data, it must be able to represent alphanumeric information (text), such as a patients name, in digital form.
Computers must also be able to represent numbers in digital form. Most computers
have several formats for numbers. For example, one computer system provides formats for 1-, 2-, 4-, and 8-byte positive integers; 1-, 2-, 4-, and 8-byte signed integers
(signed meaning that they may have positive or negative values); and 4- and 8-byte
floating-point numbers (to be discussed shortly).
When numbers can assume very large or very small values or must be represented
with great accuracy, formats requiring a large amount of storage per number must be
106
used. However, if numbers are restricted to integers within a limited range of values,
considerable savings in storage and processing time may be realized. For instance, the
grayscale values in ultrasound images typically range from 0 (black) to 255 (white),
a range of 256 numbers. As will be seen in the following section, any grayscale value
within this range can be represented by only eight bits. The following sections describe
schemes for the digital representation of different types of information.
As discussed above, computer memory and storage consist of many bits. Each
bit can be in one of two states and can therefore represent the numbers 0 or 1. Two
bits have four possible configurations (00, 01, 10, or 11) that in decimal form are 0,
1, 2, and 3. Three bits have eight possible configurations (000, 001, 010, 011, 100,
101, 110, or 111) that can represent the decimal numbers 0, 1, 2, 3, 4, 5, 6, and 7.
In general, N bits have 2N possible configurations and can represent integers from 0
to 2N 1. One byte can therefore store integers from 0 to 255 and a 16-bit word can
store integers from 0 to 65,535 (Table 5-4).
The previous discussion dealt only with positive integers. It is often necessary for
computers to manipulate integers that can have positive or negative values. There are
many ways to represent signed integers in binary form. The simplest method is to
reserve the first bit of a number for the sign of the number. Setting the first bit to 0 can
indicate that the number is positive, whereas setting it to 1 indicates that the number
is negative. For an 8-bit number, 11111111 (binary) 127 (decimal) is the most
negative number and 01111111 (binary) 127 (decimal) positive number. Most computers use a different scheme called twos complement notation to represent signed
integers, which simplifies the circuitry needed to add positive and negative integers.
Computers used for scientific purposes must be able to manipulate very large
numbers, such as Avogadros number (6.022 3 1023 molecules per gram-mole) and
very small numbers, such as the mass of a proton (1.673 3 10227 kg). These numbers are usually represented in floating-point form. Floating-point form is similar to
scientific notation, in which a number is expressed as a decimal fraction times ten
raised to a power. A number can be also written as a binary fraction times two to a
power. For example, Avogadros number can be written as
0.11111111 (binary) 201001111 ( binary ).
When a computer stores this number in floating-point form, it stores the pair of
signed binary integers, 11111111 and 01001111.
It is often necessary for computers to store and manipulate alphanumeric data, such
as a patients name or the text of this book. A common method for representing alphanumeric data in binary form has been the American Standard Code for Information InterTABLE 5-4 NUMBER OF BITS REQUIRED TO STORE INTEGERS
NUMBER
OF BITS
POSSIBLE
CONFIGURATIONS
NUMBER OF
CONFIGURATIONS
REPRESENT INTEGERS
(DECIMAL FORM)
0,1
0,1
00,01,10,11
0,1,2,3
000,001,010,011,100,101,110,111
0,1,2,3,4,5,6,7
00000000 to 11111111
256
0 to 255
16
0000000000000000 to
1111111111111111
65,536
0 to 65,535
2N
0 to 2N 1
107
change (ASCII). Each character is stored in seven bits. The byte values from 00000000
to 01111111 (binary) represent 128 characters, including the upper- and lowercase
English letters, the integers 0 through 9, many punctuation marks, and several carriagecontrol characters such as line feed. For example, the uppercase letter A is represented
by 01000001, the comma is represented by 00111010, and the digit 2 is represented
by 00110010. ASCII is being superseded by text encoding schemes that permit more
than 128 characters to be represented in digital form. Unicode is a system for encoding
the characters of the worlds languages and other symbols. Unicode incorporates ASCII.
In Unicode, as many as four bytes may be needed to represent a character.
Transfers of Information in Digital Form

Information is transferred between the various components of a computer, such as the
memory and central processing unit, in binary form. A voltage of fixed value (such as 5 V)
and fixed duration on a wire can be used to represent the binary number 1 and a voltage
of 0 V for the same duration can represent 0. A group of such voltage pulses can be used
to represent a number, an alphanumeric character, or other unit of information. (This
scheme is called unipolar digital encoding. Many other digital encoding schemes exist.)
A group of wires used to transfer data between several devices is called a data bus.
Each device connected to the bus is identified by an address or a range of addresses.
Only one device at a time can transmit information on a bus, and in most cases only
one device receives the information. The sending device transmits both the information and the address of the device that is intended to receive the information.
Advantages and Disadvantages of Analog and Digital Forms

There is a major disadvantage to the electronic transmission of information in analog
form the signals become distorted, causing a loss of fidelity of the information. Causes
of this distortion include inaccuracies when signals are amplified, attenuation losses, and
electronic noise small stray voltages that exist on circuits and become superimposed on
the signal. The more the information is transferred, the more distorted it becomes. On
the other hand, information stored or transferred in digital form is remarkably immune
to the accumulation of errors because of signal distortion. These distortions are seldom
of sufficient amplitude cause a 0 to be mistaken for a 1 or vice versa. Furthermore, most
digital circuits do not amplify the incoming information, but make fresh copies of it, thus
preventing distortions from accumulating during multiple transfers.
The digital form facilitates other safeguards. Additional redundant information can
be sent with each group of bits to permit the receiving device to detect errors or even
correct them. A simple error detection method uses parity bits. An additional bit is
transmitted with each group of bits, typically with each byte. The bit value designates
whether an even or an odd number of bits were in the 1 state. The receiving device
determines the parity and compares it with the received parity bit. If the parity of the
data and the parity bit do not match, an odd number of bits in the group have errors.
There are advantages to analog form. Information can often be transferred in less
time using the analog form. However, digital circuits are likely to be less expensive.
Conversion Between Analog and Digital Forms

Conversion of Data from Analog to Digital Form
The transducers, sensors, or detectors of most electronic measuring equipment, including medical imaging devices, produce analog data. The data must be converted to digital
form if they are to be processed by a computer, transferred over a network, or stored
108
on a digital storage device. Devices that perform this function are called analog-to-digital
converters (ADCs, also A/Ds). ADCs are essential components of all medical imaging systems producing digital images and of multichannel analyzers, discussed in Chapter 17.
The conversion of an analog signal into a digital signal is called digitization.
There are two steps in digitizationsampling and quantization. Most analog signals
are continuous in time, meaning that at every point in time the signal has a value.
However, it is not possible to convert the analog signal to a digital signal at every
point in time. Instead, certain points in time must be selected at which the conversion is to be performed. This process is called sampling. Each analog sample is then
converted into a digital signal. This conversion is called quantization.
An ADC is characterized by its sampling rate and the number of bits of output
it provides. The sampling rate is the number of times a second that it can sample
and digitize an input signal. Most radiologic applications require very high sampling
rates. An ADC produces a digital signal of a fixed number of bits. For example,
an ADC may produce an 8-bit, a 10-bit, a 12-bit, a 14-bit, or even a 16-bit digital
signal. The number of bits of output is just the number of bits in the digital number
produced each time the ADC samples and quantizes the input analog signal.
As discussed above, the digital representation of data is superior to analog representation in its resistance to the accumulation of errors. However, there are also
disadvantages to digital representation, an important one being that the conversion
of an analog signal to digital form causes a loss of information. This loss is due to both
sampling and quantization. Because an ADC samples the input signal, the values of
the analog signal between the moments of sampling are lost. If the sampling rate of
the ADC is sufficiently rapid that the analog signal being digitized varies only slightly
during the intervals between sampling, the sampling error will be small. There is a
minimum sampling rate requirement, the Nyquist limit (discussed in Chapter 4) that
ensures the accurate representation of a signal.
Quantization also causes a loss of information. As mentioned previously, an analog signal is continuous in magnitude, meaning that it can have any of an infinite
number of values between a minimum and a maximum. For example, an analog
voltage signal may be 1.0, 2.5, or 1.7893 V. In contrast, a digital signal is limited to
a finite number of possible values, determined by the number of bits used for the
signal. As was shown earlier in this chapter, a 1-bit digital signal is limited to two
values, a 2-bit signal is limited to four values, and an N-bit signal is restricted to 2N
possible values. The quantization error is similar to the error introduced when a
number is rounded off. Table 5-5 lists the maximal percent errors associated with
digital signals of various numbers of bits.
There are additional sources of error in analog-to-digital conversion other than
the sampling and quantization effects described above. For example, some averaging
TABLE 5-5MAXIMAL ERRORS WHEN DIFFERENT NUMBERS OF BITS ARE
USED TO APPROXIMATE AN ANALOG SIGNAL

NUMBER OF BITS
NUMBER OF VALUES
MAXIMAL QUANTIZATION ERROR (%)
25
12.5
256
12
4,096
6.2
0.20
0.012
109
of the analog signal occurs at the time of sampling, and there are inaccuracies in the
quantization process. In summary, a digital signal can only approximate the value of
an analog signal, causing a loss of information during analog-to-digital conversion.
No analog signal is a perfect representation of the quantity being measured.
Statistical effects in the measurement process and stray electronic voltages (noise)
cause every analog signal to have some uncertainty associated with it. To convert an
analog signal to digital form without a significant loss of information content, the
ADC must sample at a sufficiently high rate and provide a sufficient number of bits
so that the error is less than the uncertainty in the analog signal being digitized. In
other words, an analog signal with a large signal-to-noise ratio (SNR) requires an ADC
providing a large number of bits to avoid reducing the SNR.
Digital-to-Analog Conversion
It is often necessary to convert a digital signal to analog form. For example, to display digital
images from a CT scanner on a display monitor, the image information must be converted
from digital form to analog voltage signals. This conversion is performed by a digital-toanalog converter (DAC). It is important to recognize that the information lost by analog-todigital conversion is not restored by sending the signal through a DAC (Fig. 5-2).
5.2 Digital Radiological Images

A digital image is a rectangular, sometimes square, array of picture elements called pixels.
In most radiological images, each pixel is represented by a single number and the image
can be stored as a matrix of these numbers. For example, in nuclear medicine planar
images, a pixel contains the number of counts detected by the corresponding portion
of the crystal of a scintillation camera. In tomographic imaging (e.g., x-ray CT and
positron emission tomography [PET]), each pixel describes a property of a corresponding volume element (voxel) in the patient. In projection imaging (e.g., radiography),
each pixel describes a property regarding the voxels along a line projected through the
patient. Figure 5-3 shows a digital image in four different pixel formats. The number
in each pixel is converted into a visible light intensity when the image is displayed on a
display monitor. Typical image formats used in radiology are listed in Table 5-5.
Imaging modalities with higher spatial resolution and larger fields of view require
more pixels per image so the image format does not degrade the resolution. In general, an image format should be selected so that the pixel size is half or less than the
size of the smallest object to be seen. In the case of a fluoroscope with a 23-cm field
of view, the 5122 pixel format would be adequate for detecting objects as small as
(1)
01
(2)
10
(3)
11
(2)
10
(1)
01
DAC
3
1 2 3 2 1
Voltage
ADC
2
1
2-bit Digital Signal Output

Analog input
Analog output
Signal
Sampling
FIGURE 5-2 Analog-to-digital conversion and digital-to-analog conversion. In this figure, a 2-bit ADC samples the input signal five times. Note that the output signal from the DAC is only an approximation of the
input signal to the ADC because the five 2-bit digital numbers produced by the ADC can only approximate the
continuously varying analog signal. More rapid sampling and digital numbers with more bits would provide a
more accurate representation.
110
FIGURE 5-3 Effect of pixel size on image qualitydigital chest image in formats of 1,0242, 642, 322, and
162 pixels (A, B, C, and D, respectively).
about a millimeter in size close to the image receptor. To detect objects half this size,
a larger format, such as 1,024 by 1,024, should be selected. When it is necessary to
depict the shape of an object, such as a microcalcification in x-ray mammography,
an image pixel matrix much larger than that needed to merely detect the object is
required. Figure 5-3 shows the degradation of spatial resolution caused by using too
small an image matrix. The penalty for using larger pixel matrices is increased storage
and processing requirements and slower transfer of images.
The largest number that can be stored in a single pixel is determined by the number
of bits or bytes used for each pixel. If 1 byte (8 bits) is used, the maximal number that
can be stored in one pixel is 255 (28 2 1). If 2 bytes (16 bits) are used, the maximal
number that can be stored is 65,535 (216 2 1). The contrast resolution provided by an
imaging modality determines the number of bits required per pixel. Therefore, imaging
modalities with higher contrast resolution require more bits per pixel. For example, the
limited contrast resolution of ultrasound usually requires only 6 or 7 bits, and so 8 bits
are commonly used for each pixel. On the other hand, x-ray CT provides high contrast
resolution and 12 bits are required to represent the full range of CT numbers. Figure 5-4
shows the degradation of contrast resolution caused by using too few bits per pixel.
Pixel size is calculated by dividing the distance between two points in the subject
being imaged by the number of pixels between these two points in the image. It is
approximately equal to the field of view of the imaging device divided by the number of
pixels across the image. For example, if a fluoroscope has a 23-cm (9-inch) field of view
and the images are acquired in a 512-by-512 format, then the approximate size of a pixel
is 23 cm/512 = 0.045 cm 5 0.45 mm for objects at the face of the image receptor.
111
FIGURE 5-4 Effect of number of bits per pixel on image qualitydigital chest image in formats of 8, 3, 2,
and 1 bits per pixel (A, B, C, and D, respectively). Too few bits per pixel not only causes loss of contrast resolution, but also creates the appearance of false contours.
The total number of bytes required to store an image is the number of pixels in
the image multiplied by the number of bytes per pixel. For example, the number of
bytes required to store a 512-by-512 pixel image, if one byte is used per pixel, is
(512 512 pixels) (1 byte/pixel) / (1,000 bytes/kB) 262 kB.
Similarly, the number of 512-by-512 images that may be stored on a 60-GB optical
disk, if 16 bits (2 bytes) are used per pixel, is
(60 GB/disk) (109 bytes/GB) / [(512 512 pixels/image) (2 bytes/pixel)]
114,000 images/disk,
if no other information is stored on the disk.
If these are CT studies and an average of 500 images per study is stored, this disk
would hold about 114,000 images/500 images per study 228 studies. In practice,
other information stored with the studies would slightly reduce the number of studies that could be stored on the disk.
5.3 Digital Computers

Computers were originally designed to perform mathematical computations
and other information processing tasks very quickly. Since then, they have come
to be used for many other purposes, including information display, information
112
storage, and, in conjunction with computer networks, information transfer and

communications.
Computers were introduced in medical imaging in the early 1970s and have
become increasingly important since that time. Today, computers are essential to
most imaging modalities, including x-ray CT, magnetic resonance imaging (MRI),
single photon emission computed tomography (SPECT), and PET.
Any function that can be performed by a computer can also be performed by a
hard-wired electronic circuit. The advantage of the computer over a hard-wired circuit is its flexibility. The function of the computer can be modified merely by changing the program that controls the computer, whereas modifying the function of a
hard-wired circuit usually requires replacing the circuit. Although the computer is a
very complicated and powerful information processing device, the actual operations
performed by a computer are very simple. The power of the computer is mainly due
to its speed in performing these operations and its ability to store large volumes of
information.
The components and functioning of computers are discussed in Appendix B. The
remainder of this section will define several terms regarding computers that will be
used in the following parts of this chapter.
The term hardware refers to the physical components of a computer or other
device, whereas software refers to the programs, consisting of sequences of instructions, that are executed by a computer. Software is commonly categorized as applications programs or systems software.
An applications program, commonly referred to as an application, is a program
that performs a specific function or functions for a user. Examples of applications
are e-mail programs, word processing programs, web browsers, and image display
programs.
A computers operating system is a fundamental program that is executing even
when a computer seems to be idle and awaiting a command. When the user instructs
a computer to run a particular program, the operating system copies the program
into memory from a disk, transfers control to it, and regains control of the computer
when the program has finished. An operating system handles many housekeeping
tasks, such as details of storage of information on disks and magnetic tapes, apportioning system resources amongst applications and amongst users in a multiuser system, and handling interrupts, for example, when a user activates a pointing device.
Examples of operating systems are Microsoft Windows and Linux on IBM compatible
PCs; OS X on Apple computers; and Unix, used on a wide variety of computers.
A workstation is a computer designed for use by a single person at a time. A workstation is usually equipped with one or more display monitors for the visual display
of information, a keyboard for the entry of alphanumeric information, and a pointing
device, such as a mouse.
5.4 Information Storage Devices

Mass storage devices permit the nonvolatile (i.e., data are not lost when power is
turned off) storage of information. Most mass storage devices have mechanical components, but at least one newer storage modality, flash memory, is entirely electronic.
Mass storage devices with mechanical components include magnetic disk drives,
magnetic tape drives, and optical (laser) disk units. Each of these devices consists of
a mechanical drive; the storage medium, which may be removable; and an electronic
controller. Despite the wide variety of storage media, there is not one best medium
113
for all purposes, because they differ in storage capacity, data access time, data transfer
rate, cost, and other factors.
When the CPU or another device sends data to memory or a mass storage
device, it is said to be writing data, and when it requests data stored in memory or
on a storage device, it is said to be reading data. Data storage devices permit either
random access or sequential access to the data. The term random access describes
a storage medium in which the locations of the data may be read or written in any
order, and sequential access describes a medium in which data storage locations can
only be accessed in a serial manner. Most solid-state memories, magnetic disks,
and optical disks are random access, whereas magnetic tape typically permits only
sequential access.
Magnetic Disks
Magnetic disks are spinning disks coated with a material that may be readily
magnetized. Close to the surface of the spinning disk is a read-write head that, to
read data, senses the magnetization of individual locations on the disk and, to write
data, changes the direction of the magnetization of individual locations on the disk.
Most disk drives have a read-write head on each side of a platter so that both sides
can be used for data storage. Information is stored on the disk in concentric rings
called tracks. The read-write heads move radially over the disk to access data on individual tracks. The access time of a disk is the time required for the read-write head to
reach the proper track (head seek time) and for the spinning of the disk to bring the
information of interest to the head (rotational latency). The data transfer rate is the
rate at which data are read from or written to the disk once the head and disk are in
the proper orientation; it is primarily determined by the rotational speed of the disk
and the density of information storage in a track.
A typical hard magnetic disk drive, as shown in Figure 5-5, has several rigid platters
stacked above each other on a common spindle, with a read-write head for each side of
each platter. Hard disks with nonremovable platters are called fixed disks. The platters
continuously rotate at a high speed (typically 5,400 to 15,000 rpm). The read-write heads
aerodynamically float at distances less than a micron above and below the disk surfaces
on air currents generated by the spinning platters. A spinning hard disk drive should
Spindle
Actuator
Read-write
head
Platters
Cable
connector
Power connector
FIGURE 5-5 Hard magnetic disk drive with 500 GB storage capacity. A read-write head is visible at the end
of the actuator arm overlying the top disk platter. There is a read-write head for each surface of each platter.
114
not be jarred because that might cause a head crash in which the head strikes the disk,
gouging the disk surface and destroying the head, with a concomitant loss of information.
The portion of a disk drive containing the disks and read-write heads is sealed to keep
dirt from damaging them. Hard disks have very large storage capacities, up to 2 TB each.
Although their access times and data transfer rates are very slow compared to solid-state
memory, they are much faster than those of most other storage media. For these reasons,
hard disks are used on most computers to store frequently used programs and data.
Flash Memory
A type of solid-state memory called flash memory stores digital information as small electric charges. Flash memory does not require electrical power to maintain stored information and thus can replace spinning magnetic disks and other storage media. Today,
flash memory is much more expensive per unit storage capacity than magnetic disks
and other storage media and so is not used commonly for the storage of large amounts
of information. However, it is replacing spinning magnetic disks on some portable laptop computers. Flash drives, which connect to the USB ports of workstations and
laptops, have mostly replaced floppy disk drives and are rapidly replacing optical (CD
and DVD) disks because of their portability and relatively large capacities.
Magnetic Tape
Magnetic tape is plastic tape coated with a magnetizable substance. Its average data
access times are very long, because the tape must be read serially from the beginning
to locate a particular item of information. Magnetic tape was originally stored on
reels, but today is obtained in cartridges or cassettes. There are several competing
cartridge or cassette formats available today. A single tape cartridge or cassette can
store a very large amount of data, up to about 1.5 TB uncompressed. Common uses
are to back up (make a copy for the purpose of safety) large amounts of important
information and archival storage of digital images.
Optical Disks
An optical disk is a removable disk that rotates during data access and from which
information is read and written using a laser. There are three categories of optical
disksread-only; write-once, read-many-times (WORM); and rewritable. Read-only
disks are encoded with data that cannot be modified. To read data, the laser illuminates a point on the surface of the spinning disk and a photodetector senses the
intensity of the reflected light. WORM devices permit any portion of the disk surface
to be written upon only once, but to be read many times. This largely limits the use
of WORM optical disks to archival storage. To store information, a laser at a high
intensity burns small holes in a layer of the disk. To read information, the laser, at a
lower intensity, illuminates a point on the spinning disk and a photodetector senses
the intensity of the reflected light. Rewritable optical disks permit the stored data to
be changed many times. Most rewritable optical disks today use phase-change technology. The recording material of a phase-change disk has the property that, if heated
to one temperature and allowed to cool, it becomes crystalline, whereas if heated
to another temperature, it cools to an amorphous phase. To write data on a phasechange disk, the laser heats a point on the recording film, changing its phase from
crystalline to amorphous or vice versa. The transparency of the amorphous material
differs from that in the crystalline phase. Above the layer of recording material is a
reflective layer. Information is read as described above for WORM disks.
115
CDs, which have capacities of about 650 MB, are available as WORM and rewritable
optical disks. The CD has been partially displaced by a newer standard for optical disks,
the DVD, that provides a much larger storage capacity because the laser beam is focused
to a smaller spot on the recording layer. DVDs are also available in WORM and rewritable
forms and provide storage capacities of about 4.5 GB. Optical disks using 405 nm wavelength blue-violet lasers are now available. The blue-violet beam can be focused onto a
smaller area because of its shorter wavelength, increasing the storage capacity substantially. There are two competing standards for these disks, both available in WORM and
rewritable formats, providing up to 50 or 60 GB per disk. Optical disks provide much
prompter data access than magnetic tape and better information security than most other
media because they are not subject to head crashes, as are magnetic disks; are not as vulnerable to wear and damage as magnetic tape; and are not affected by magnetic fields.
Table 5-6 compares the characteristics of mass storage devices and memory. Because
there is no best device or medium for all purposes, most computer systems today have
at least a hard magnetic disk drive and a drive capable of reading optical disks.
Technologies for Large Archives

Medical imaging can produce very large amounts of information. For example, a medium
size medical center with a radiology department, cardiac catheterization laboratory, and
several services performing fluoroscopically guided interventional procedures typically
produces several terabytes of image information each year. There are technologies that
permit automated storage and retrieval of such massive amounts of information.
TABLE 5-6COMPARISON OF CHARACTERISTICS OF MASS STORAGE MEDIA
AND MEMORY
STORAGE
REMOVABLE CAPACITY
ACCESS TIME TRANSFER

(AVERAGE)
RATE
COST PER MEDIA

DISK/TAPE COST PER GB
Hard magnetic
disk
Usually not
20 GB2 TB 615 ms
3170 MB/s NA
$0.05
Solid state
disk (flash
memory)
Yes or no
20 GB2 TB ,0.1 ms
Up to
740 MB/s
NA
$2.00
Optical disk,
CD-R, CD-RW
Yes
usually
650 MB
3.6 MB/s
(for 24)
$0.25,
$1.00
$0.40,
$1.50
Optical disk,
DVD-R,
DVD-RAM
Yes
36 GB
$0.20,
$2.00
$0.04,
$0.40
Optical disk,
blue-violet
laser
Yes
2360 GB
25 ms
12 MB/s
$60
(UDO2)
$1.00
(UDO2)
Magnetic tape
(cartridge or
cassette)
Yes
45 MB1.5
TB
Seconds to
minutes
0.125140
MB/s
$85 (1.5
TB LTO-5
$0.05 (1.5
TB LTO-5)
DRAM solidstate memory
No
NA
180 ns
0.53 GB/s
NA
NA
100150 ms
(for 24)
26 MB/s
(for 20)
Note: Values are typical for 2011. Cost refers to one disk or tape cartridge or tape cassette. Storage capacities
and transfer rates are for uncompressed data; they would be higher for compressed data.
ms, milliseconds; ns, nanoseconds (106 ns 1 ms); MB, megabytes; GB, gigabytes; TB, terabytes; LTO, linear tape
open. Prices and capabilities of enterprise quality (high reliability and long lifetime) disks are ,10 higher
than commodity grade disks.
116
A technology called RAID (redundant array of independent disks) can provide a large
amount of on-line storage. RAID permits several or many small and inexpensive hard
magnetic disk drives to be linked together to function as a single very large drive. There
are several implementations of RAID, designated as RAID Levels 0, 1, 2, 3, 4, 5, and 10.
In RAID Level 0, portions of each file stored are written simultaneously on several disk
drives, with different portions on each disk drive. This produces very fast reads and writes,
but with no redundancy, and so the loss of one of these drives results in the loss of the file.
In RAID Level 1, called disk mirroring, all information is written simultaneously onto two
or more drives. Because duplicate copies of each file are stored on each of the mirrored
disk drives, Level 1 provides excellent data protection, but without any improvement in
data transfer rate and with at least a doubling of required storage capacity. The other RAID
levels provide various compromises among fault tolerance, data transfer rate, and storage
capacity. Most RAID implementations provide sufficient redundancy that no information
is lost by the failure of a single disk drive. Figure 5-6 shows RAID modules.
Another technology permitting the storage of very large amounts of information is
automated magnetic tape or optical disk libraries, commonly referred to as jukeboxes.
In these devices, magnetic tape cartridges or cassettes or optical disks are stored in
large racks. Robotic arms load the media into or remove them from drives that read
and write information on the media. These automated libraries commonly have two
or more drives, permitting simultaneous access to more than one unit of media. The
storage capacity of a jukebox is the product of the storage capacity of a unit of media
and the number of media units that can be stored on the racks. The access time for a
study stored on media in a jukebox includes a few seconds for the disk, cartridge, or
cassette to be loaded into a drive as well as the access time for the media.
5.5 Display of Digital Images

As mentioned above, an important function of many computers, particularly workstations, is to display information. In medical imaging, the purpose of the display may be
to permit technologists to visually assess the adequacy of acquired images, for physicians to interpret images, or to guide physicians performing interventional procedures.
The designs of display systems should take into account the human visual system.
Single RAID module with controller
Stacked RAID modules (6) in rack mount
FIGURE 5-6 Stackable RAID modules. Each module contains a dual CPU processor, two 140 GB disks for
the CPU and software, and eight 300 GB disk drives in a RAID-5 configuration, with one hot-swappable drive.
Total storage is 2 TB per module, 1.6 of which are usable for data storage (0.8 TB used for redundancy). The 6
stacked modules provide about 10 TB of uncompressed data storage (1/2 rack) with dimensions of approximately
2 3 4 ft (w d h).
117
The display system of a workstation or other computer typically consists of a

display interface, also referred to by terms such as video interface and graphics
adapter, in the computer; one or more display monitors connected to the display
interface by a cable or cables that carry electrical signals; and software to control
the display system. The display interface and display monitor(s) may be consumergrade commercial products or they may be more expensive devices designed for the
high-fidelity display of medical images.
A display system may be designed to display color images or only monochrome
(grayscale) images. A system that can display color images can also display grayscale
images. Nearly all general-purpose commercial display systems are designed to display
images in color. However, images from the radiological modalities, consisting of rectangular arrays of numbers, in general do not benefit from being displayed in color and
so are commonly displayed on high-quality grayscale monitors for interpretation. However, some radiological images, such as co-registered dual modality images (e.g., SPECT/
CT and PET/CT), false color images (commonly used in nuclear medicine myocardial
perfusion imaging), and volume rendered displays (discussed later in this chapter) of
three-dimensional image sets, benefit from being displayed partly or entirely in color.
Color is a perception created by some animals visual systems. In the retina of the
human eye, cone cells provide color vision. In most humans, there are three types of
cone cells, which differ in spectral sensitivity. One is most sensitive to yellow-green light
(although largely responsible for the perception of red), another to green light, and the
third to blue light. A mixture of red, green, and blue light, in the proper proportion,
can cause the impression of many colors and of white. In a color display monitor, each
displayed pixel generates red, green, and blue light, with the intensities of each independently controlled. The construction of display monitors is described below.
In grayscale radiological images, each pixel value is typically represented by a
single integer, commonly stored in 16 bits, although fewer than 16 of the bits may be
used. For example, the range of CT numbers requires only 12 bits. In color images,
each pixel is commonly represented by three integers, commonly 8 or 10 bits each,
which are contained in a 32-bit word. The three integers designate the intensities of
the red, green, and blue light to be generated for that pixel.
Display Interface and Conversion of a Digital Image into

Analog Signals for Display
A computers display interface converts a digital image into a signal that is displayed by a
display monitor. Most display interfaces contain memories to store the images being displayed. For a digital image to be displayed, it is first sent to this display memory by the computers CPU under the control of an application program. Once it is stored in this memory,
the display interface of the computer reads the pixels in the image sequentially and sends
these values to the monitor for display. Older display interfaces produce analog signals that
are displayed by monitors; such a display interface contains one or more digital-to-analog
converters (DACs), which convert each digital number to an analog voltage signal. Newer
display interfaces send the image information in digital form to the display monitors; these
monitors contain DACs to convert the digital pixel values to analog signals.
Display Monitors
For approximately 80 years, cathode ray tubes (CRTs) were used for the electronic
display of images. CRT displays are rapidly being replaced by flat-panel displays and
can no longer be obtained on most medical imaging equipment. There are several
118
types of flat-panel displays, including gas plasma, light emitting diode, organic light
emitting diode, and field emissive displays. However, most flat-panel monitors used
today to display medical images have liquid crystal displays (LCDs) because of their
superior performance and longevity.
The term pixel, when referring to display monitors, has a meaning different
from a picture element of an image; it refers to a physical structure in a monitor
that displays a single pixel of an image. Most types of monitors have physical pixels,
although CRT monitors do not.
Cathode Ray Tube Monitors

A CRT in a display monitor is an evacuated container, usually made of glass, with
components that generate, focus, and modulate the intensity of one or more electron
beams directed onto a fluorescent screen. When a voltage is applied between a pair
of electrodes, the negative electrode is called the cathode and the positive electrode is
called the anode. If electrons are released from the cathode, the electric field accelerates them toward the anode. These electrons are called cathode rays. A grayscale CRT
monitor (Fig. 5-7) generates a single electron beam. The source of the electrons is
a cathode that is heated by an electrical current and the anode is a thin aluminum
layer in contact with the fluorescent screen. A large constant voltage, typically 10 to
30 kV, applied between these two electrodes creates an electric field that accelerates
the electrons to high kinetic energies and directs them onto the screen. The electron
beam is focused, electrically or magnetically, onto a very small area on the screen.
Between the cathode and the screen is a grid electrode. A voltage applied to the grid
electrode is used to vary the electric current (electrons per second) of the beam. The
electrons deposit their energy in the phosphor of the screen, causing the emission of
visible light. The intensity of the light from any location on the screen is proportional
to the electric current in the beam, which is determined by the analog voltage signal
applied to the grid electrode. The electron beam is steered, either electrically or magnetically, in a raster pattern as shown in Figure 5-8.
The CRT of a color monitor is similar to that of a grayscale monitor, but has three
electron guns instead of one, producing three electron beams. The three electron beams
FIGURE 5-7 Grayscale CRT display monitor.

The electron gun contains a heated cathode, the
source of the electrons in the beam, and a grid
electrode, which modulates the intensity (electrical current) of the beam. This CRT uses an electromagnet to focus the beam and two pairs of
electromagnets to steer the beam. (One pair, not
shown, is perpendicular to the pair in the diagram.) Alternatively, electrodes placed inside the
CRT can perform these functions.
119
FIGURE 5-8 Order of conversion of pixels

in a digital image to analog voltages to form
a video signal (raster scan). Data is scanned
from left to right, with a retrace to the next
row (dashed line). The actual size of the matrix
is much smaller than depicted.
are close to one another and are steered as one, but their intensities are modulated
separately. The fluorescent screen consists of triads of tiny dots or stripes of phosphors
emitting red, green, and blue light. Just before the screen is a thin metal sheet containing holes or slits so that each electron beam only strikes the phosphor emitting a single
color. Thus one electron beam produces red light, another produces green light, and
the third produces blue light. As mentioned above, mixtures of red, green, and blue
light can create the perception of many colors by the human visual system.
Liquid Crystal Display Monitors

In LCDs, the pixels are physical structures; in many grayscale LCDs and all color
LCDs, each pixel consists of three subpixels, whose light intensities are independently controlled. The liquid crystal (LC) material of an LCD does not produce light.
Instead, it modulates the intensity of light from another source. All high image quality LCDs are illuminated from behind (backlit). A backlit LCD consists of a uniform
light source, typically containing fluorescent tubes or light emitting diodes and a
layer of diffusing material; a layer of LC material between two glass plates; and a
polarizing filter on each side of the glass holding the LC material (Fig. 5-9).
Visible light, like any electromagnetic radiation, may be considered to consist of
oscillating electrical and magnetic fields, as described in Chapter 2. The oscillations
are perpendicular to the direction of travel of the light. Unpolarized light consists of
light waves whose oscillations are randomly oriented. A polarizing filter is a layer of
material that permits components of light waves oscillating in one direction to pass,
but absorbs components oscillating in the perpendicular direction. When unpolarized light is incident on a single layer of polarizing material, the intensity of the transmitted light is half that of the incident light. When a second polarizing filter is placed
in the beam of polarized light, parallel to the first filter, the intensity of the beam
transmitted through the second filter depends on the orientation of the second filter
with respect to the first. If both filters have the same orientation, the intensity of the
beam transmitted through both filters is almost the same as that transmitted through
the first. However, if the second filter is rotated with respect to the first, the intensity
of the transmitted light is reduced. When the second filter is oriented so that its
polarization is perpendicular to that of the first filter, almost no light is transmitted.
120
Unpolarized
fluorescent
light source
Horizontal
Polarizing
Filter
LC
layer
Off
Vertical
Polarizing
Filter
Pixel
Output
Dark
Bright
Horizontally
Polarized
Light
On
Vertically
Polarized
Light
FIGURE 5-9 Two subpixels of a grayscale backlit LCD monitor. On each side of the LC layer in each subpixel
is a transparent electrode. In the top subpixel, no voltage is applied across the LC layer. The polarization of the
light is unchanged as it passes through the LC layer, causing most of the light to be absorbed by the second
polarizing filter and thereby producing a dark subpixel. In the bottom subpixel, a voltage is applied across the
LC layer. The applied voltage causes the molecules in the LC layer to twist. The twisted molecules change the
polarization of the light, enabling it to pass through the second filter with an intensity that increases with the
applied voltage; with full voltage on the LC layer, the output light is vertically polarized, causing the subpixel
to transmit the brightest luminance. (The bars shown in the polarizing filters are merely an artists rendition to
indicate the direction of the polarization, as is depicting the polarized light as a ribbon.) Variations in grayscale
are achieved by varying the voltage provided to each LCD subpixel.
An LC material consists of long organic molecules and has properties of both a

liquid and a crystalline material. For example, it flows like a liquid. On the other
hand, the molecules tend to align parallel to one another. The material has additional
properties that are used in LCDs. If a layer of LC material is in contact with a surface
with fine grooves, the molecules align with the grooves. If an electric field is present,
the molecules will align with the field. If polarized light passes through a layer of LC
material, the LC material can change the polarization of the light.
In an LCD display, the light that has passed through the first filter is polarized.
Next, the light passes through a thin layer of LC material contained between two
glass plates. The sides of the glass plates in contact with the LC material have fine
parallel grooves to orient the LC molecules. Each pixel or subpixel of an LCD has a
pair of electrodes. When a voltage is applied to the pair of electrodes, an electric field
is created, changing the orientation of the LC molecules and the polarization of the
light. The amount of change in the orientation of the molecules and the polarization
of the light increases with the magnitude of the applied voltage. The light then passes
through the second polarizing filter, which can be oriented so that pixels or subpixels
are bright when no voltage is applied, or so that they are black when no voltage is
applied. Figure 5-9 shows the filters oriented so that the subpixels are dark when no
voltage is applied. Thus, when no voltage is applied to a pixel or subpixel, the light is
polarized by the first filter, passes through the LC material, and then is absorbed by the
second polarizing filter, resulting in a dark pixel or subpixel. As the voltage applied to
121
the pixel or subpixel is increased, the LC molecules twist, changing the polarization of
the light and decreasing the fraction absorbed by the second filter, thereby making the
pixel or subpixel brighter.
Three major variants of LCD technology are twisted nematic (TN), in-plane
switching (IPS), and patterned vertical alignment (PVA). The differences among
these designs, discussed in some of the references listed at the end of this chapter,
are mainly due to the positions of the two electrodes in each subpixel and in the orientations of the LC molecules without applied voltage and when voltage is applied.
The performance of these designs is more important to the user. TN LCDs, commonly used in inexpensive home and office monitors, suffer from very limited viewing angles and thus should be avoided when purchasing medical image displays.
More expensive IPS panels have significantly improved viewing angles and excellent
color reproduction. PVA was implemented as a compromise between the TN and IPS
technologies, with an intermediate price.
A color LCD has an additional layer containing color filters. Each pixel
consists of three subpixels, one containing a filter transmitting only red light, the
second transmitting only green light, and the third transmitting only blue light.
As mentioned previously, mixtures of red, green, and blue light can create the
perception of most colors. Because these color filters absorb light, they reduce the
luminance of the display, in comparison to an equivalent monochrome LCD, by
about a factor of three.
In theory, each pixel or subpixel of a flat-panel display could be controlled by
its own electrical conductor. If this were done, a three-megapixel grayscale display
would contain at least 3 million electrical conductors and a color display of the same
pixel format would have three times as many. In practice, flat-panel displays are
matrix controlled, with one conductor serving each row of pixels and one serving
each column. For a three-megapixel grayscale display (2,048 by 1,536 pixels), only
2,048 row pathways and 1,536 column pathways (if there are not subpixels) are
required. A signal is sent to a specific pixel by simultaneously providing voltages to
the row conductor and the column conductor for that pixel.
The intensity of each pixel must be maintained while signals are sent to other pixels. In active matrix LCDs, each pixel or subpixel has a transistor and capacitor. The
electrical charge stored on the capacitor maintains the voltage signal for the pixel or
subpixel while signals are sent to other pixels. The transistors and capacitors are constructed on a sheet of glass or quartz coated with silicon. This sheet is incorporated as
a layer within the LCD. Active matrix LCDs are also called thin film transistor (TFT)
displays. TFT technology, without the polarizing filters and LC material, is used in
flat-panel x-ray image receptors and is discussed in Chapter 7.
Performance of Display Monitors

Display monitors usually are the final component of the imaging chain and their performance can significantly affect radiological imaging. Monitors are characterized by
parameters such as spatial resolution, spatial distortion, contrast resolution, aspect
ratio, maximal luminance, black level, dynamic range, uniformity of luminance,
noise, lag, and refresh rate.
The photometric quantity* describing the brightness of a monitor (or other light
source) is luminance. Luminance is the rate of light energy emitted or reflected from a
*Photometric quantities and units describe the energy per unit time carried by light, modified to account
for the spectral sensitivity of the human eye. A person with normal vision perceives a given radiance of
green light as being brighter than, for example, an equal radiance of red or blue light.
122
surface per unit area, per unit solid angle, corrected for the photopic spectral sensitivity
of the human eye. The SI unit of luminance is the candela per square meter (cd/m2).
Perceived brightness is not proportional to luminance; for example, the human visual
system will perceive a doubling of luminance as only a small increase in brightness.
The contrast resolution of a monitor is mainly determined by its dynamic range,
defined as the difference between its maximal and minimal luminance. (Sometimes,
the dynamic range is defined as the ratio of the maximal to minimal luminance.)
Studies have been undertaken with light boxes and display monitors on the effect
of maximal luminance on diagnostic performance (Goo et al, 2004; Krupinski, 2006).
Studies involving light boxes and some of those involving display monitors showed
better accuracy with higher luminance. However, other studies using display monitors,
particularly those in which contrast adjustment (e.g., windowing and leveling) by the
observer was permitted, showed little or no effect of maximal luminance on diagnostic
performance. At least one study showed that lower luminance increased the time to
reach diagnostic decisions. A likely explanation is that contrast adjustment can, at least
partially, compensate for the adverse effect of lower maximal monitor luminance.
Especially bright monitors should be used for the display of radiological images
to provide adequate dynamic range. The American College of Radiology has published the ACR Technical Standard for Electronic Practice of Medical Imaging; the current
standard specifies that grayscale (monochrome) monitors of interpretation workstations should provide maximal luminances of at least 171 cd/m2, whereas those for
mammography should provide at least 250 cd/m2 and, for optimal contrast, at least
450 cd/m2. Grayscale monitors for displaying medical images for interpretation can
typically provide maximal luminances of about 600 to 900 cd/m2 when new and are
usually calibrated to provide maximal luminances of about 400 to 500 cd/m2 that can
be sustained for several tens of thousands of operating hours. Grayscale LCD monitors provide about three times larger maximal luminances than do color LCD monitors with backlights of equal intensity; this is because the red, green, and blue filters
in color LCDs absorb light to create color images. Color LCD monitors providing
sustained luminances exceeding 400 cd/m2 are available; however, they are not able
to sustain these luminances as long as comparable grayscale monitors. To provide
the same maximal luminance, an LCD with smaller pixels must have a brighter backlight than one with larger pixels. The stability of the luminance of an LCD monitor
depends upon the constancy of the luminance of the backlight. Most LCD monitors
designed for medical imaging have light sensors that are used to keep the luminance
constant; in some cases, the sensor is in the back of the monitor and in other cases,
it is in front and measures the luminance seen by the viewer.
The minimal luminance of a monitor, measured in total darkness with the entire
screen black, is called the black level. LCD monitors have much higher black levels
(the luminance is higher) than the CRT monitors they have replaced. The contrast
ratio is the maximal luminance divided by the black level. In practice, when viewing
clinical images in a room that is not completely dark, the minimal luminance achievable is determined not only by the black level, but also by the scattering of ambient
light from the face of the monitor and by veiling glare. Veiling glare is stray light
from the face of the monitor that occurs when an image is displayed. The amount
of veiling glare at a specific location on the monitors face is determined by the size
and brightness of other areas of the displayed image and the distances from those
areas. Veiling glare causes a reduction in image contrast. LCD monitors suffer less
The word photopic refers to the normal daylight color vision of the human visual system. The photopic spectral sensitivity differs from the scotopic spectral sensitivity of dark-adapted night vision.
123
from veiling glare than do CRT monitors. In practice, monitors are not viewed in
completely dark rooms. The diffuse reflection of ambient light from the face of the
monitor reduces image contrast. The specular (mirror-like) reflection of light from
bright objects imposes structured noise on the displayed image. The faces of LCD
monitors reflect less ambient light than do those of CRT monitors.
Spatial linearity (freedom from spatial distortion) describes how accurately shapes
and lines are presented on the monitor. Because the pixel matrix is physically built
into LCD monitors, they provide excellent spatial linearity.
CRT monitors, whose images are formed by scanning electron beams, must refresh
the image at a rate greater than a threshold to avoid the perception of flicker in the image.
This threshold, called the flicker-fusion frequency, increases with the brightness of the
image and the fraction of the visual field occupied by the image and varies from person
to person; it ranges from about 60 to over 80 frames per second for very bright monitors.
Active matrix LCD monitors, unlike CRT monitors, do not exhibit flicker when displaying stationary images and so the refresh rate is irrelevant in this situation. However, the
refresh rate of an LCD monitor does matter when displaying dynamic images. The frame
refresh rate needed to provide the appearance of continuous motion when viewing cine
images is less than that needed to avoid the perception of flicker and may be as low as
about 25 frames per second. LCD monitors typically suffer more from lag in displaying
dynamic images than do CRT monitors. The lag in LCD monitors is caused by the time
required to change the electrical charges stored by the small capacitors in individual
pixels. Monitors add both spatial and temporal noise to displayed images.
There are practical differences between LCD and CRT monitors. LCD monitors
require much less space, are much lighter, consume less electrical power, and generate
much less heat. A problem of LCD monitors is nonfunctional pixels or subpixels; these
may be permanently off or permanently on. The fraction of nonfunctional pixels or
subpixels should be small and they should not be grouped together. Another disadvantage of LCD monitors is limited viewing angle, which varies with the LCD technology.
The luminance and apparent contrast are reduced if the monitor is not viewed from
nearly directly in front. In distinction, CRT monitors permit very wide viewing angles.
The lifetimes of LCD monitors, limited by the brightness of the fluorescent tubes in the
backlights, exceed those of CRT monitors, limited by the electron guns and phosphor.
The spatial resolution of a display monitor is primarily described by its addressable
pixel format (e.g., 1,280 by 1,024 pixels) with respect to the useful display area. An
active matrix LCD monitor provides spatial resolution superior to a CRT monitor with
the same addressable pixel format. Because the individual pixels are physical structures,
LCD monitors provide uniform spatial resolution over their entire faces. Improvements
in the performance of LCD monitors have resulted in their replacing CRT monitors.
The necessary dimensions of the active face of a display monitor and its pixel format (number of rows and columns of pixels or, equivalently, the pixel pitch, defined as
the distance from the center of a pixel to the center of an adjacent pixel in the same row
or column) depend upon the pixel matrix sizes of the images produced by the modalities whose images will be displayed, the distance from the viewers eyes to the display
surface, and the spatial resolution of the human visual system. Digital mammograms
and radiographs contain many more pixels than other radiological images (Table 5-7)
and are referred to as large matrix-size images.
It is desirable for a radiologist to view an entire image at or near maximal spatial
resolution. To avoid a reduction in spatial resolution when an image is displayed, a
monitor should have at least as many pixels in the horizontal and vertical directions as
The standard for motion pictures in the United States is 24 frames per second.
124
TABLE 5-7TYPICAL RADIOLOGIC IMAGE FORMATS

MODALITY
PIXEL FORMAT
BITS PER PIXEL
Scintillation camera planar
642 or 1282
8 or 16
SPECT
642 or 1282
PET
128 to 336
16
Digital fluoroscopy
5122 or 1,0242
8 to 12
Fuji CR chest (200 m)
2,140 3 1,760
10 to 12
Trixell DR (143 m)
3,000 3 3,000
12 to 16
GE24 3 31 cm (100 m)
2,394 3 3,062
1216
Hologic24 3 29 cm (70 m)
3,328 3 4,096
X-ray CT
5122
12
MRI
642 to 1,0242
12
Ultrasound
512
8 or 16
2
Digital radiography
Mammography
CT, computed tomography; MRI, magnetic resonance imaging; PET, positron emission tomography; SPECT,
single photon emission computed tomography.
the image. On the other hand, it serves no purpose to provide spatial resolution that is
beyond the ability of the viewer to discern. The distance from a persons eyes to the face
of a display monitor of a workstation commonly ranges from about 50 to 60 cm. The
viewer may lean closer to the monitor for short periods to see more detail, but viewing
at a distance much closer than about 55 cm for long periods is uncomfortable.
To assess perceived contrast and spatial resolution, scientists studying human
vision commonly use test images containing a sinusoidal luminance pattern centered
in an image of constant luminance equal to the average luminance of the sinusoidal
pattern (Fig. 5-10). Contrast may be defined as
C = (Lmax 2 Lmin)/Lavg,
where Lmax is the maximal luminance in the pattern, Lmin is the least luminance, and Lavg
is the average luminance. The smallest luminance difference (Lmax 2 Lmin) that is detectable by half of a group of human observers is known as a just noticeable difference (JND).
The threshold contrast is the JND divided by Lavg. Contrast sensitivity is defined as
FIGURE 5-10 Sinusoidal luminance
test patterns are used by researchers to
assess the perception of contrast as a
function of spatial frequency and average
luminance. A. Four different spatial frequencies with the background luminance
equal to the average luminance of the
pattern are shown. B. Four different
contrast levels at a fixed frequency. Sinusoidal lines (in blue) above the patterns
show the change in luminance as amplitude and frequency variations
Frequency variations at fixed contrast
Contrast variations at fixed spatial frequency
Contrast Sensitivity
400
300
125
FIGURE 5-11 Contrast sensitivity of the human

visual system as a function of spatial frequency.
Adapted from Flynn MG. Visual requirements for
high-fidelity display. In: Advances in digital radiography: categorical course in diagnostic radiology physics. Oak Brook, IL: Radiological Society
of North America, 2003:103107.
60 cm viewing distance
100 cd / m2 Luminance
21 mm2 test pattern
200
100
0
0.01
0.1
10
Spatial Frequency, mm1
the inverse of the threshold contrast. Figure 5-11 is a graph of contrast sensitivity as
a function of spatial frequency (cycles per mm at a 60 cm viewing distance) from a
typical experiment. Studies have shown that people with good vision perceive image
contrast best at about five cycles per visual degree. The threshold contrast is reduced
to about a tenth of the maximum at about 20 cycles per visual degree and reduced
to less than one hundredth of the maximum at about 40 to 50 cycles per visual
degree (Barten, 1999). At a 60-cm viewing distance, a visual degree is equivalent
to a distance of 10.5 mm (10.5 mm 5 60 cm 3 tan 1) on the face of the monitor.
Thus, at a 60-cm viewing distance, a person with good vision perceives contrast best
at about half a cycle per mm, perceives contrast reduced to less than a tenth of this at
about two cycles per mm, and perceives very little contrast beyond about four cycles
per mm. Two cycles per mm is approximately equivalent to 4 pixels per mm, or a
pixel pitch of 1 mm per 4 pixels 5 0.25 mm and four cycles per mm is approximately
equivalent to 8 pixels per mm, or a pixel pitch of 1 mm per 8 pixels 0.125 mm.
Monitor Pixel Formats

Display monitors are available in many face sizes and pixel formats. Two common formats for commercial color monitors are 1,280 by 1,024 pixels and 1,600 by 1,200 pixels. Digital radiographs are typically stored in formats of about 2,000 by 2,500 pixels. A
common active display size for monitors used to display radiographs, including mammograms, is 54 cm (21 inches) diagonal. The faces of the monitors typically have aspect
ratios (width to height) of about 3 to 4 or 4 to 5. A common pixel format for these monitors is 2,560 pixels by 2,048 pixels; such monitors are commonly called 5 megapixel
monitors. This pixel format permits the display of an entire large matrix image at near
maximal resolution. Less-expensive 54-cm diagonal monitors have pixel formats of
2,048 by 1,536 pixels, commonly called 3 megapixel monitors, and 1,6003 1,200
pixels, commonly called 2 megapixel monitors. A typical pixel pitch of a 5-megapixel
monitor is 0.165 mm and that of a 3-megapixel monitor is 0.21 mm.
LCD monitors with 76 cm (30 inches) diagonal faces are available in ten-, 6-, and
4-megapixel formats. Such a monitor is intended to replace a pair of 54-cm diagonal
monitors and permit the simultaneous side-by-side display of two radiographs or
mammograms in full or near full spatial resolution.
A monitor orientated so that the horizontal field is greater than the vertical is
called landscape display, whereas an orientation with the vertical field greater than
the horizontal is called portrait display. Monitors used for the display of digital
radiographs are usually used in the portrait orientation, although the newer 76 cm
diagonal monitors that display two radiographs in full or near full resolution sideby-side are used in the landscape orientation.
126
When an entire image is displayed on a display monitor with a smaller number

of horizontal or vertical pixels than the image itself, the image must be reduced to
a smaller pixel format by averaging neighboring pixels in the stored image, causing
a loss of spatial resolution. The zoom function permits a portion of the image to be
displayed in full spatial resolution on such a monitor and the pan function on the
workstation allows the viewing of different portions of the image.
Display of Image Contrast and Contrast Enhancement

A display system must be able to display a range of pixel values from the minimal
pixel value to the maximal value in the image. For displaying the image, a range of
light intensities is available, from the darkest to the brightest the display monitor
is calibrated to produce. A great amount of choice exists in how the mapping from
pixel value to light intensity is performed. A mapping can be selected that optimizes
the contrast of important features in the image, thereby increasing their conspicuity.
Alternatively, if this mapping is poorly chosen, the conspicuity of important features
in the image can be reduced.
Lookup tables (LUTs) are commonly used by medical image processing and display
computers to affect the display of image contrast. Such use of a LUT may be considered
to be a method of image processing. However, it is intimately connected with displaying images and so is discussed here. A LUT is simply a table containing a value for each
possible pixel value in an image. For example, if each pixel in an image could have
one of 4,096 values, a LUT would have 4,096 elements. In practice, each pixel value in
the transformed image is determined by selecting the value in the LUT corresponding
to the pixel value in the unmodified image. For example, if the value of a pixel in the
unmodified image is 1,342, the value of the corresponding pixel in the transformed
image is the 1,342nd value in the LUT. Figure 5-12 illustrates the use of a LUT.
A LUT may be applied to transform image contrast at more than one point in the
chain from image acquisition to image display. For example, a digital radiography system may use a LUT to modify the acquired pixel values, which are proportional to the
detected x-ray signals, to cause the display of contrast to resemble that of a film (Fig.
5-13). Similarly, a medical grade image display system may employ a LUT to modify each
pixel value before it is sent to the DAC and becomes an analog signal; this is commonly
done to compensate for the different display functions of individual display monitors
and is described below in this chapter. Figure 5-13 shows five LUTs in graphical form.
Digital Look-Up-Table (LUT)
255
Output
from LUT
DAC
Input to LUT
Raw digital Image Data
0
4095
Enhanced Analog
Image at Dispay
FIGURE 5-12 Display interface showing function of a lookup table (LUT). An image is transferred to the
memory of the display interface. The display interface selects pixel values in a raster pattern, and sends the
pixel values, one at a time, to the LUT. The LUT produces a digital value indicating display intensity to the DAC.
The DAC converts the display intensity from a digital value to an analog form (e.g., a voltage).

Defaut Modality LUT
Linear LUT
VOI LUT
Inverted LUT
W=16384 L=8192
W=3674 L=4384
127
High Contrast LUT
Output
Bright
Drak
0
input
16384
W=16384 L=8192
W=3674 L=4384
W=2048 L=5600
FIGURE 5-13 Graphs of five digital lookup tables (LUT), for the same 14 bit digital image and corresponding
window and level settings. From left to right, the first is a linear LUT that preserves the way the image
was originally acquired. Typically, the useful image data occupies only a small range of values for this 14 bit
image, and thus contrast is low. The second is the default modality LUT that is assigned by the modality,
based upon an optimized grayscale range. Note that a large fraction of the range of the image is set to zero
(dark) or largest output (maximal brightness), and that a small range of values is mapped from the dark to
the bright values on the display. The third is the Value of Interest LUT (VOILUT), which encompasses the
full range of input values and maps the output according to the capabilities of the display. Typically, this LUT
is a sigmoidally-shaped curve, which softens the appearance of the image in the dark and bright regions. The
fourth LUT inverts image contrast. Shown here is the inverted second image. The fifth image demonstrates
windowing to enhance contrast in underpenetrated parts of the image by increasing the slope of the LUT.
Note that this causes the loss of all contrast in the highly penetrated regions of the lung.
Windowing and leveling is a common method for contrast enhancement. It permits

the viewer to use the entire range of display intensities to display just a portion of the
total range of pixel values. For example, an image may have a minimal pixel value of
0 and a maximum of 200. If the user wishes to enhance contrast differences in the
brighter portions of the image, a window from 100 to 200 might be selected. Then,
pixels with values from 0 to 100 are displayed at the darkest intensity and so will not
be visible on the monitor. Pixels with values of 200 will be displayed at the maximal
brightness, and the pixels with values between 100 and 200 will be assigned intensities
determined by the current LUT. Figure 5-13 (rightmost image) illustrates windowing.
Most display workstations have level and window controls. The level control
determines the midpoint of the pixel values to be displayed and the window control
determines the range of pixel values about the level to be displayed. Some nuclear
medicine workstations require the user to select the lower level (pixel value below
which all pixels are displayed at the darkest intensity) and the upper level (pixel
value above which all pixel values are set to the brightest intensity):
Lower level 5 Level 2 Window/2
Upper level 5 Level 1 Window/2.
Windowing and leveling can be performed with a LUT or, equivalently, by calculation; if calculated, it is accomplished by subtraction of a number from each pixel,
followed by multiplication by another number. If calculated, each pixel value equal
to or below the lower level is set to zero and each equal to or above the upper level
is set to the maximal pixel value possible. For each pixel whose value is between the
lower and upper level, the windowed pixel value is calculated as
Pi,j = (Pij Lower level) (maximal allowed pixel value/window),
128
VeriLUM
Calibration Luminance Responce Plot
Head: 0
245.66
Luminance
73.18
21.80
6.50
016 032 048 064 080 096 112 128 144 160 176 192 208 224 240 255
+ Luminance Responce Model
* Characteristic Curve Plus Ambient Offset
1.94
Driving Level
FIGURE 5-14 A typical response function (digital driving level versus output luminance) of an LCD monitor
is shown as the white x symbols. Note the nonlinear luminance response as a function of digital driving
level. The black + symbols are the monitor luminance values that conform to the DICOM Grayscale Standard
Display Function (GSDF), PS 3.14. For the noncalibrated monitor, the lower luminance areas of the image will
be displayed suboptimally. (See Fig. 5-18 for a comparison of calibrated and uncalibrated monitors.) The Luminance Response Model in black is a portion of the DICOM GSDF curve shown in Figure 5-16.
where Pij is a pixel value before windowing, Pi,j is the pixel value after windowing,
maximal pixel value is the largest possible pixel value, and window upper level
lower level.
Luminance Calibration of Display Systems

The display function of a display monitor describes the luminance produced as a function
of the magnitude of the digital (sometimes called a digital driving level) or analog
signal sent to the monitor. The display function of a monitor can greatly affect the
perceived contrast of the displayed image. The inherent display function of a monitor
is nonlinear (increasing the input signal does not, in general, cause a proportional
increase in the luminance), varies from monitor to monitor, and also changes with time.
Figure 5-14 shows a display function of an LCD monitor (white curve). Furthermore,
individual display monitors differ in maximal and minimal luminance.
A display monitor has controls labeled brightness and contrast that are used
to adjust the shape of the display function. On most monitors used for medical image
interpretation, they are not available to the user.
After acquisition of digital images by an imaging device, the pixel values may be
adjusted automatically by the devices computer so that the displayed images have an
appearance acceptable to radiologists (black curve in Fig. 5-14). For example, digital
radiography systems, including digital mammography systems, modify the pixel values to give the images appearances similar to screen-film radiographs. Other processing may be performed, such as the automatic contrast equalization processing of some
digital mammography systems (Chapter 8), to increase the conspicuity of structures of
interest and to reduce the need for the interpreting physician to window and level the
images. When these images are sent to a workstation for viewing, unless the display
functions of the display devices were known to the designer of the imaging device, the
images contrast intended by the imaging device designer may not be achieved. Also,
the images appearance may also be modified by a technologist using a review workstation prior to transfer to the PACS. If the display system of the review workstation
P-Values
User adjustment
of contrast
129
Standardized Display System

DDLs
P-Values
to DDLs
FIGURE 5-15 This diagram shows the use of a lookup table (LUT), labeled P-values to DDLs, with each display
system, such as a display monitor, so that the net display function, provided by the LUT and display device, will
conform to the DICOM Grayscale Standard Display Function (GSDF). The DICOM GSDF does not replace adjustment of contrast by the user. (Adapted from Digital Imaging and Communications in Medicine (DICOM) Part 14:
Grayscale Standard Display Function, PS 3.14-2011, National Electrical Manufacturers Association, 2011.)
has a display function significantly different from that of the interpretation workstation, the interpreting physician may see contrast displayed quite differently than the
technologist intended. Furthermore, if the monitors on display stations have different display functions, an image may have a different appearance on different display
monitors, even when more than one monitor is attached to a single workstation.
To resolve this problem, the Digital Imaging and Communications in Medicine
(DICOM) Grayscale Standard Display Function (GSDF) was created to standardize
the display of image contrast. The DICOM GSDF pertains to grayscale (monochrome)
images, whether displayed on monitors or printed onto paper or onto film for viewing on a light box. It does not apply to color images, although it does pertain to
grayscale images displayed by a color monitor. The goals of the DICOM GSDF are to
1. Provide applications with a predictable transformation of digital pixel values
(called presentation values, abbreviated as P-values) to luminance
2. Insofar as possible, provide a similar display of contrast on display monitors and
printed media
3. Insofar as possible, provide perceptual linearization, that is, equal differences in the
pixel values received by the display system should be perceived as equal by the
human visual system
The first two of these goals are met by the DICOM GSDF. In fact, on display monitors
having identical minimal and maximal luminances, the DICOM GSDF will provide
almost identical displays of contrast. The third goal, perceptual linearization, can
likely not be perfectly achieved, given the complexity of the human visual system;
nonetheless, the DICOM GSDF comes close to achieving it.
The display function of a monitor can be modified to any desired shape by the
use of a LUT in the workstation or monitor, as described in earlier in this chapter.
The LUT contains an output pixel value for each possible pixel value in an image. In
the DICOM standards, the input pixel values provided to the LUT are called presentation values, as shown in Figure 5-15, and the output values of the LUT that
are provided to the display system are called digital driving levels. Thus, by using
a LUT for each monitor, the net display function (monitor and LUT) will conform to
the DICOM GSDF.
Figure 5-16 shows the DICOM GSDF response model, which specifies luminance as a function of a parameter called JND index; JND is an acronym for
just-noticeable difference, discussed earlier in this section. DICOM PS 3-14, 2011,
defines the JND index as The input value to the GSDF, such that one step in JND
The DICOM GSDF is part of the DICOM standards discussed later in this chapter.
130
DICOM Grayscale Standard Display Function

10000
Luminance (Cd / m2)
1000
Laser Flim
100
CRT, LCD Displays
10
1
0.1
0.01
100
200
300
400 500 600

JND Index
700
800
900 1000 1100
FIGURE 5-16 The DICOM Grayscale Standard Display Function. Only a segment of the curve from the
minimum to the maximum luminance values of a display device, such as a CRT or LCD monitor or laser printed
film, is used as the calibrated display function. For instance, the arrows point to the operating range of the LCD
monitor whose response is shown in Figure 5-14. The maximum operating ranges for CRT or LCD displays and
laser film in terms of luminance relative to the JND index are shown in red and blue highlights, respectively. A
calibrated monitor or film response will fall along a subset of the ranges depicted above.
Index results in a Luminance difference that is a Just-Noticeable Difference. The

shape of the GSDF is based on a model of the human visual system. A detailed
explanation of this model is beyond the scope of this book; it is discussed in DICOM
PS 3-14, 2011, and Flynn 2003.
Only a portion of the DICOM GSDF curve, that lies between the minimal and
maximal calibrated luminances of a monitor, is used as the display function for that
monitor. For example, as shown in Figure 5-16, if the minimal and maximal calibrated luminances of a display monitor are 2 and 250 cd/m2, the portion of the
DICOM GSDF contained in the segment encompassing 2 and 250 cd/m2 is used as
the display function. The presentation values p, from 0 to the maximal possible presentation value (typically 2N 1, where N is the number of bits used for a presentation value), are linearly related to the values of JND index j for that segment:
j = jmin 1 p (jmax 2 jmin)/pmax,
where jmin is the smallest JND index in the rectangle, jmax is the largest JND index in
the rectangle, and pmax is the maximum possible presentation value. Thus, a presentation value of 0 is assigned the lowest luminance of the monitor, which is the lowest luminance in the part of the GSDF within the segment. The maximum possible
presentation value pmax is assigned the maximal calibrated luminance of the monitor,
which is the maximal luminance in the segment. Each intermediate presentation
value p is assigned the luminance on the GSDF corresponding to the JND index j
determined from the equation above.
The calibration of a display monitor to the DICOM GSDF is usually performed
automatically by the workstation itself, using specialized software and a calibrated
photometer that sends a digital luminance signal to the workstation. The photometer
is aimed at a single point on the face of the monitor (Fig. 5-17). The software, starting with a digital driving level of zero, increases the digital driving levels provided to
the display system in a stepwise fashion and the photometer sends to the software
the measured luminance for each step, thereby measuring the display function of the
131
External
Internal
Stored
Deployed
FIGURE 5-17 A display monitor can be calibrated so its corrected display function conforms to the DICOM
GSDF. This is accomplished by using a photometer that measures the luminance at the center of the monitor
as a program varies the digital driving levels that control the luminance of pixels in a square area at the center
of the monitor. The photometer sends the digitized luminance values to the computer. The program then
calculates and stores values in a LUT to cause the monitor to conform to the GSDF. The upper picture shows a
monitor with a photometer manually placed at the center of the monitor. The lower picture shows a medicalgrade monitor with a built-in internal photometer that calibrates the display automatically.
display system. The software then calculates the values to be placed in the LUT that
will cause the LUT and display system, acting together, to conform to the DICOM
GSDF. Some display monitors are equipped with photometers and can automatically
assess monitor luminance and GSDF calibration (Fig. 5-17, bottom.).
The DICOM GSDF also can be applied to printed images, whether on transmissive media (film on a viewbox) or reflective media (e.g., an image printed on paper).
Shown in Figure 5-18 is an image of a Briggs contrast test pattern for a display
without luminance calibration (left) and luminance calibration conforming to the
DICOM GSDF (right).
False Color Displays

As mentioned above, radiographic images do not inherently have the property of color.
When color is used to display them, they are called false color images. Figure5-19 shows
how false color images are produced by nuclear medicine computer workstations. Common uses of false color displays include nuclear medicine, in which color is often used to
enhance the perception of contrast, and ultrasound, in which color is used to superimpose flow information on images displaying anatomic information. False color displays
132
FIGURE 5-18 A Briggs contrast

pattern. Left. The uncalibrated monitor display (its uncorrected display
function is shown in Fig. 5-14 as the
white symbols) shows the difficulty
in resolving the contrast differences
in the darker areas of the image
(highlighted red box). Right. The
calibrated monitor display (response
to luminance variations is graphed
in Fig. 5-14 as the black symbols)
shows improved contrast response.
Uncalibrated
CONTINUE
Calibrated
GCT disabled
CONTINUE
GCT enabled
are also commonly used in image co-registration, in which an image obtained from one
modality is superimposed on a spatially congruent image from another modality. For
example, it is often useful to superimpose nuclear medicine images, which display physiology, but may lack anatomic landmarks, on CT or MRI images. In this case, the nuclear
medicine image is usually superimposed in color on a grayscale MRI or CT image.
Performance of Diagnostic Monitors versus That

of Consumer-Grade Monitors
Radiological images are displayed for interpretation using very expensive LCD monitors designed for medical image display instead of consumer-grade LCD monitors.
Advantages of these diagnostic monitors commonly include
1. Much higher maximal luminance
2. More uniform luminance
3. Smaller pixels, providing higher spatial resolution
4. Wider viewing angles, due to different LCD technology
Digital signal
Analog signal
DAC
Red
Translation
table
Digital signal
Analog signal
DAC
Digital
image
Green
Translation
table
Video display memory

in video interface
to green subpixel
of video monitor
Analog signal
Digital signal
DAC
Blue
Translation
table
to red subpixel
of video monitor
to blue subpixel
of video monitor
FIGURE 5-19 Creation of a false color display in nuclear medicine. Each individual pixel value in the image
being displayed is used to look up a red, a green, and a blue intensity value. These are simultaneously displayed
adjacent to one another within a single pixel of a color display monitor. The mix of these three colors creates
the perception of a single color for that pixel.
133
5. More than 8 bits for digital driving levels (grayscale monitors) or more than 8bits
each for the digital driving levels for red, green, and blue (color monitors), to
display subtle contrast differences
5. Automatic luminance stabilization circuits to compensate for changing backlight
intensity
6. Hardware and software for implementing the DICOM GSDF
7. Ability to monitor display monitor luminance and GSDF calibration over a network
Cameras to Record Digital Images on Film

Despite the increasing capability to transfer and view medical images without the use
of photographic film, there are still situations in which digital images must be recorded
on film. For example, federal mammography regulations in effect require each facility
performing full-field digital mammography to have such a capability. Today, images
are commonly recorded on film by multiformat laser cameras, also known as laser
imagers. The images are usually sent to the device by a computer network. These
devices usually provide several formats (e.g., one image, four images, and six images
per sheet). After exposure of the film, it is chemically developed and viewed on a light
box. Although films with silver halide grains and wet chemical processing were once
used, nearly all laser imagers today use silver-based film and a dry thermal development process to convert the latent images on the exposed film into visible images.
A laser camera typically contains a microprocessor; an image memory; an analog-todigital converter; a laser with a modulator to vary the intensity of the beam; a rotating
polygonal mirror to scan the laser beam across the film; and a film-transport mechanism
to move the film in a linear fashion so the scanning covers the entire film. Each sweep of
the laser beam across the film records a single row of pixel values. The lasers in most laser
cameras emit red or infrared light, requiring the use of red- or infrared-sensitive film.
Care must be taken not to handle this film under the normal red darkroom safelight.
5.6 Computer Networks
Definitions and Basic Principles

Computer networks permit the transfer of information between computers, allowing
computers to share devices, such as printers, laser multiformat cameras, and information storage devices, and enabling services such as electronic transmission of messages (e-mail), transfer of computer files, and use of distant computers. Networks,
based upon the distances they span, may be described as local area networks (LANs)
or wide area networks (WANs). A LAN connects computers within a department, a
building such as medical center, and perhaps neighboring buildings, whereas a WAN
connects computers at large distances from each other. LANs and WANs evolved
separately and a computer can be connected to a WAN without being part of a LAN.
However, most WANs today consist of multiple LANs connected by medium or long
distance communication links. The largest WAN is the Internet.
A server is a computer on a network that provides a service to other computers on
the network. A computer with a large array of magnetic disks that provides data storage for other computers is called a file server. There are also print servers, application
servers, database servers, e-mail servers, web servers, etc. A computer, typically a
workstation, on a network that makes use of a server is called a client.
Two common terms used to describe client-server relationships are thick client
and thin client. Thick client describes the situation in which the client computer
134
rovides most information processing and the function of the server is mainly to store
p
information, whereas thin client describes the situation in which most information
processing is provided by the server and the client mainly serves to display the information. An example would be the production of volume rendered images from a set of CT
images. In the thin client relationship, the volume rendered images would be produced
by the server and sent to a workstation for display, whereas in a thick client relationship,the images would be produced by software and/or a graphics processor installed
on the workstation. The thin client relationship can allow the use of less capable and
less expensive workstations and enable specialized software and hardware, such as a
graphics processor, on a single server to be used by several or many workstations.
Networks have both hardware and software components. A physical connection must exist between computers so that they can exchange information. Common
physical connections include coaxial cable, copper wiring, optical fiber cables, and
radio, including microwave, communication systems. A coaxial cable is a cylindrical cable with a central conductor surrounded by an insulator that, in turn, is surrounded by a tubular grounded shield conductor. Coaxial cable and copper wiring
carry electrical signals. Optical fiber cables use glass or plastic fibers to carry nearinfrared radiation signals produced by lasers or light emitting diodes. Optical fiber
cables have several advantages over cables or wiring carrying electrical signals, particularly when the cables must span long distancesoptical fiber is not affected by
electrical interference and thus usually has lower error rates; it can carry signals
greater distances before attenuation, distortion, and noise require the use of repeaters
to read and retransmit the signals; and it permits much greater information transmission rates. There are also layers of software between the application program with
which the user interacts and the hardware of the communications link.
Network protocols are standards for communication over a network. A protocol
provides a specific set of services. Both hardware and software must comply with
established protocols to achieve successful transfer of information. Failure to conform to a common protocol would be analogous to a person speaking only English
attempting to converse with a person speaking only Arabic. Several hardware and
software protocols will be described later in this chapter.
In most networks, multiple computers share communication pathways. Most
network protocols facilitate this sharing by dividing the information to be transmitted into packets. Packets may be called frames, datagrams, or cells. Some protocols
permit packets of variable size, whereas others permit only packets of a fixed size.
Each packet has a header containing information identifying its destination. In most
protocols, a packet also identifies the sender and contains information used to determine if the packets contents were garbled during transfer. The final destination computer reassembles the information from the packets and may request retransmission
of lost packets or packets with errors.
Large networks usually employ switching devices to forward packets between
network segments or even between entire networks. Each device on a network,
whether a computer or switching device, is called a node and the communications
pathways between them are called links. Each computer is connected to a network
by a network adapter, also called a network interface, installed on the I/O bus of the
computer or incorporated on the motherboard. Each interface between a node and
a network is identified by a unique number called a network address. A computer
usually has only a single interface, but a switching device connecting two or more
networks may have an address on each network.
The maximal data transfer rate of a link or a connection is called the bandwidth, a
term originally used to describe the data transfer capacities of analog communications
135
channels. An actual network may not achieve its full nominal bandwidth because of
overhead or inefficiencies in its implementation. The term throughput is commonly
used to describe the maximal data transfer rate that is actually achieved. Bandwidth
and throughput are usually described in units of megabits per second (106 bps 5
1 Mbps) or gigabits per second (109 bps 5 1 Gbps). These units should not be
confused with megabytes per second (MBps) and gigabytes per second (GBps),
commonly used to specify the data transfer rates of computer components such as
magnetic and optical disks. (Recall that a byte consists of eight bits. Megabytes and
gigabytes are defined in Table 5-3.) The latency is the time delay of a transmission
between two nodes. In a packet-switched network, it is the time required for a small
packet to be transferred. It is determined by factors such as the total lengths of the
links between the two nodes, the speeds of the signals, and the delays caused by any
intermediate repeaters and packet switching devices.
On some small LANs, all computers are directly connected, that is, all packets
reach every computer. However, on a larger network, every packet reaching every
computer would be likely to cause unacceptable network congestion. For this reason,
most networks larger than a small LAN, and even many small LANs, employ packet
switching. The packets are sent over the network. Devices such as switches and routers
(to be discussed shortly) store the packets, read the destination addresses, and send
them on toward their destinations, a method called store and forward. In some very
large packet-switched networks, individual packets from one computer to another
may follow different paths through the network and may arrive out-of-order.
Networks are commonly designed in layers, each layer following a specific protocol. Figure 5-20 shows a model of a network consisting of five layers. Each layer in
the stack provides a service to the layer above.
The top layer in the stack is the Application Layer (Layer 5 in Fig. 5-20). Application programs, commonly called applications, function at this layer. Applications
are programs that perform useful tasks and are distinguished from systems software,
such as an operating system. On a workstation, applications include the programs,
such as an e-mail program, word processing program, web browser, or a program
for displaying medical images, with which the user directly interacts. On a server,
an application is a program providing a service to other computers on the network.
The purpose of a computer network is to allow applications on different computers
to exchange information.
Network communications begin at the Application Layer on a computer. The application passes the information to be transmitted to the next lower layer in the stack.
Layer5
Application
Application
Layer4
Transport
Transport
Layer3
Network
Network
Layer2
Data Link
Data Link
Layer1
Physical
network
Physical
FIGURE 5-20 TCP/IP Network Protocol Stack. Another network model,

the Open Systems Interconnection
(OSI) model, is also commonly used
to model networks. The OSI model
has seven layers. The bottom four
layers of the OSI model match the
bottom four layers shown in this
figure. The OSI model has two additional layers between the transport
layer and the application layer.
136
The information is passed from layer to layer, with each layer adding information,
such as addresses and error-detection information, until it reaches the Physical Layer
(Layer 1 in Fig. 5-20). The Physical Layer sends the information to the destination
computer, where it is passed up the layer stack to the application layer of the destination computer. As the information is passed up the layer stack on the destination
computer, each layer removes the information appended by the corresponding layer
on the sending computer until the information sent by the application on the sending
computer is delivered to the intended application on the receiving computer.
The lower network layers (Layers 1 and 2 in Fig. 5-20) are responsible for transmission of packets from one node to another over a LAN or point-to-point link, and
enable computers with dissimilar hardware and operating systems to be physically
connected. The lowest layer, the Physical Layer, transmits physical signals over a
communication channel (e.g., the copper wiring, optical fiber cable, or radio link
connecting nodes). The protocol followed by this layer describes the signals (e.g.,
voltages, near-infrared signals, or radiowaves) sent between the nodes. Layer 2, the
Data Link Layer, encapsulates the information received from the layer above into
packets for transmission across the LAN or point-to-point link, and transfers them
to Layer 1 for transmission. The protocol followed by Layer 2 describes the format
of packets sent across a LAN or point-to-point link. It also describes functions such
as media access control (determining when a node may transmit a packet on a LAN)
and error checking of packets received over a LAN or point-to-point link. It is usually
implemented in hardware.
Between the lower layers in the protocol stack and the Application Layer are
intermediate layers that mediate between applications and the network interface.
These layers are usually implemented in software and incorporated in a computers
operating system. Many intermediate level protocols are available, their complexity depending upon the scope and complexity of the networks they are designed
to serve. Some functions of these intermediate layers will be discussed later in this
chapter when the methods for linking multiple LANs are described.
Local Area Networks

LAN protocols are typically designed to permit the connection of computers over
limited distances. On some small LANs, the computers are all directly connected
and so only one computer can transmit at a time and usually only a single computer
accepts the information. This places a practical limit on the number of computers
and other devices that can be placed on a LAN without excessive network congestion. The congestion caused by too many devices on a LAN can be relieved by dividing the LAN into segments connected by packet switching devices, such as bridges,
switches, and routers, that only transmit information intended for other segments.
Most types of LANs are configured in bus, ring, or star topologies (Fig. 5-21). In a
bus topology, all of the nodes are connected to a single cable. One node sends a packet,
including the network address of the destination node, and the destination node
accepts the packet. All other nodes must wait. In a ring topology, each node is only
directly connected to two adjacent nodes, but all of the nodes and their links form a
ring. In a star topology, all nodes are connected to a central node, often called a hub.
Ethernet
The most commonly used LAN protocols are the various forms of Ethernet. Before
transmission over Ethernet, information to be transmitted is divided into packets,
each with a header specifying the addresses of the transmitting and destination
Bus Topology
Ring Topology
137
Star Topology
FIGURE 5-21 Network topologiesbus, ring and star. Early forms of Ethernet used the bus topology; only
one computer at a time could transmit packets. Other networks have used the ring topology. Modern forms
of Ethernet use the star topology. The central node of the star is usually a packet switch that enables multiple
pairs of nodes to simultaneously exchange packets. Multiple stars can be connected by connecting the central
nodes.
nodes. Ethernet is contention based. A node ready to transmit a packet first listens
to determine if another node is transmitting. If none is, it attempts to transmit. If
two nodes inadvertently attempt to transmit at nearly the same moment, a collision
occurs. In Ethernet, collisions are normal expected events. When a collision occurs,
each node ceases transmission, waits a randomly determined but traffic-dependent
time interval, and again attempts to transmit. This media access control protocol is
called carrier-sense multiple-access with collision detection (CSMA/CD).
Ethernet has evolved greatly since its invention, resulting in several varieties of
Ethernet (Table 5-8). In early forms of Ethernet, only a single node could transmit
at a time on a LAN segment and usually only one node would receive and store the
packets. All other nodes had to wait. Modern forms of Ethernet are configured in a
star topology with a switch as the central node. The switch does not broadcast the
TABLE 5-8 COMMON LAN STANDARDS

TYPE
Switched Ethernet
(10 Base-T)
BANDWIDTH PHYSICAL
MEDIUM
(MBPS)
TOPOLOGY
MAXIMAL RANGE
10
star
UTP (common
100 m from node to
telephone wiring) switch
100
star
Category 5 UTP
100 m from node to

switch
Gigabit Ethernet
(1000 Base-CX and
1000 Base-T)
1,000
star
copper wiring
at least 25 m and
perhaps
100 m
Gigabit Ethernet
(1000 Base-SX and
1000 Base-LX)
1,000
star
optical fiber cable 550 m (multimode cable)

and 3 km (single mode)
100
ring
optical fiber cable 2 km (multimode) and

40 km (single mode)
between each pair of
nodes
Switched Fast
Ethernet (100
Base-TX)
FDDI
Note: Early forms of Ethernet, which used the bus topology, are not listed. FDDI has largely been replaced by
the various forms of Ethernet. In all but FDDI, the bandwidth is between each pair of nodes, with multiple
nodes being able to simultaneously transmit. In FDDI, only a single node can transmit at a time.
138
packets to all nodes. Instead, it stores each packet in memory, reads the address
on the packet, and then forwards the packet only to the destination node. Thus,
the switch permits several pairs of nodes to simultaneously communicate at the full
bandwidth (either 10 or 100 Mbps) of the network. 10Base-T Ethernet supports data
transfer rates up to 10 Mbps. Fast Ethernet (100Base-TX) is like 10Base-T, except that
it permits data transfer rates up to 100 Mbps over high-grade Category 5 UTP wiring.
There are also optical fiber versions of Ethernet and Fast Ethernet. Gigabit Ethernet
and Ten Gigabit Ethernet provide bandwidths of one and ten Gbps, respectively.
Other LAN Protocols

Many LAN protocols other than Ethernet have achieved wide implementation in
the past. These include Token Ring, AppleTalk, and Fiber Distributed Data Interface
(FDDI). These protocols have been largely replaced by the many forms of Ethernet.
Extended Local Area Networks

An extended LAN connects facilities, such as the various buildings of a medical center, over a larger area than can be served by a single LAN segment. An extended LAN
is formed by connecting individual LAN segments. This is commonly performed
by devices such as packet switches that only transmit packets when the addresses
of the transmitting and destination nodes are on opposite sides of the device. Such
a device operates at the Data Link Layer (Layer 2) of the network protocol stack.
It stores each incoming packet in memory, reads its LAN destination address, and
then forwards the packet toward its destination. Thus, in addition to extending a
LAN, these devices segment it, thereby reducing congestion. Links, sometimes called
backbones, of high bandwidth media such as Gigabit or Ten Gigabit Ethernet, may
be used to carry heavy information traffic between individual LAN segments.
Large Networks and Linking Separate Networks

It is often desirable to connect separate networks, which may use different hardware
and software protocols. A network of networks is called an internetwork or internet
(with a lower-case i). Today, most WANs are internets formed by linking multiple
LANs (Fig. 5-22). There are several problems encountered when constructing an
Router
West Wherever
Medical Center
T1 Line
T3 Line
Router
East Bogaloosa
Outpatient Clinic
Router
St. Barney
Medical Center
FIGURE 5-22 Wide area networks are commonly formed by linking together two or more local area
etworks using routers and links (e.g., T1 and T3 lines) leased from a telecommunications company.
n
139
internet. One is the routing of packets across the internet. LAN protocols are not
designed for efficient routing of packets across very large networks. Furthermore, the
addresses of nodes on a LAN, even though they may be unique, are not known to the
nodes on other LANs. Additional problems are posed by the dissimilar protocols that
may be used in individual LANs and point-to-point links.
WANs and separate networks are usually connected by devices called routers
and long distance links provided by telecommunications companies, as shown in
Figure 5-22. Routers are specialized computers or switches designed to route packets
among networks. Routers perform packet switchingthey store incoming packets,
determine the intended destinations, and, by following directions in routing tables,
forward the packets toward their destinations. Each packet may be sent through several routers before reaching its destination. Routers communicate among each other
to determine optimal routes for packets.
Routers follow a protocol that assigns each interface in the connected networks
a unique network address distinct from its LAN address. Routers operate at the Network Layer (Layer 3 in Fig. 5-20) of the network protocol stack. The dominant
routable protocol today is the Internet Protocol (IP), described below. The function
of a router is described in greater detail below.
The Internet Protocol SuiteTCP/IP

The Internet Protocol Suite, commonly called TCP/IP, is a packet-based suite of protocols used by many large networks and the Internet. TCP/IP permits information
to be transmitted from one computer to another across a series of networks connected by routers. TCP/IP is specifically designed for internetworking, that is, linking separate networks that may use dissimilar lower-level protocols. TCP/IP operates
at protocol layers above those of lower layer protocols such as Ethernet. The two
main protocols of TCP/IP are the Transmission Control Protocol (TCP), operating at the
Transport Layer (Layer 4 in Fig. 5-20) and the Internet Protocol (IP), operating at the
Network Layer (Layer 3 in Fig. 5-20).
Communication begins when an application (at Layer 5 in Fig. 5-20) passes
information to the Transport Layer, along with information designating the destination computer and the application on the destination computer, which is to
receive the information. The Transport Layer, following TCP, divides the information into packets, attaches to each packet a header containing information
such as a packet sequence number and error-detection information, and passes
the packets to the Network Layer. The Network Layer, following IP, may further subdivide the packets. The Network Layer adds a header to each packet
containing information such as the source address and the destination address.
The Network Layer then passes these packets to the Data Link Layer (Layer 2 in
Fig. 5-20) for transmission across the LAN or point-to-point link to which the
computer is connected.
The Data Link Layer, following the protocol of the specific LAN or point-to-point
link, encapsulates the IP packets into packets for transmission over the LAN or
point-to-point link. Each packet is given another header containing information
such as the LAN address of the destination computer. For example, if the lower level
protocol is Ethernet, the Data Link Layer encapsulates each packet it receives from
the Network Layer into an Ethernet packet. The Data Link Layer then passes the
packets to the Physical Layer, where they are converted into electrical, infrared, or
radio signals and transmitted.
Each computer and router is assigned an IP address. Under IP Version 4 (IPv4),
an IP address consists of four 1-byte binary numbers, permitting over 4 billion
140
distinct addresses. The first part of the address identifies the network and the latter
part identifies the individual computer on the network. A new version of IP, IPv6,
designed to ultimately replace IPv4, permits a much larger number of addresses.
IP addresses do not have meaning to the lower network layers. IP defines methods
by which a sending computer determines, for the destination IP address, the next
lower layer address, such as a LAN address, to which the packets are to be sent by
the lower network layers.
IP is referred to as a connectionless protocol or a best effort protocol. This
means that the packets are routed across the networks to the destination computer
following IP, but some may be lost enroute. IP does not guarantee delivery or even
require verification of delivery. On the other hand, TCP is a connection-oriented
protocol providing reliable delivery. Following TCP, Network Layer 4 of the sending computer initiates a dialog with Layer 4 of the destination computer, negotiating matters such as packet size. Layer 4 on the destination computer requests
the retransmission of any missing or corrupted packets, places the packets in the
correct order, recovers the information from the packets, and passes it up to the
proper application.
The advantages of designing networks in layers should now be apparent. LANs
conforming to a variety of protocols can be linked into a single internet by installing a
router in each LAN and connecting the routers with point-to-point links. The pointto-point links between the LANs can also conform to multiple protocols. All that is
necessary is that all computers and routers implement the same WAN protocols at
the middle network layers. A LAN can be replaced by one conforming to another
protocol without replacing the software in the operating system that implements
TCP/IP and without modifying application programs. A programmer developing an
application need not be concerned with details of the lower network layers. TCP/IP
can evolve without requiring changes to applications programs or LANs. All that is
necessary is that each network layer conforms to a standard in communicating with
the layer above and the layer below.
Routers
The packet switching performed by a router is different from that performed by
Ethernet switches, which merely forward identical copies of received packets. On a
LAN, the packets addressed to the router are those intended for transmission outside
the LAN. The LAN destination address on a packet received by the router is that of
the router itself.
A router performs the functions of the three lowest network levels shown in
Figure 5-20. On the router, Layer 1 receives the signals forming a packet and
passes the packet to Layer 2. Layer 2 removes the WAN packet from the lower
level protocol packet in which it is encapsulated and passes it to Layer 3. Layer
3 reads the WAN address (in WANs conforming to the Internet Protocol Suite,
an IP address) and, using a routing table, determines the optimal route for forwarding the packet. If the link for the next hop of the packet is part of a LAN,
the router determines the LAN address of the next node to which the packet is
to be sent. Layer 3 then sends the WAN packet and the LAN address of the next
node to Layer 2, which encapsulates the WAN packet into a packet conforming
to the lower level protocol for the next hop. Layer 2 passes this packet to Layer 1,
which transmits the packet on toward its ultimate destination. Thus, the packets
transmitted by a router may conform to a different lower level protocol than those
received by the router.
141
The Internet
The Internet (with a capital letter I) is an international network of networks using
the TCP/IP protocol. A network using TCP/IP within a single company or organization is sometimes called an intranet.
The Internet is not owned by any single company or nation. The main part of
the Internet consists of national and international backbone networks, consisting
mainly of fiber optic links connected by routers, provided by major telecommunications companies. These backbone networks are interconnected by routers. Large
organizations can contract for connections from their networks directly to the backbone networks. Individual people and small organizations connect to the Internet by
contracting with companies called Internet service providers (ISPs), which operate
regional networks that are connected to the Internet backbone networks.
IP addresses, customarily written as four numbers separated by periods (e.g.,
152.79.110.12), are inconvenient. Instead, people use host names, such as www.
ucdmc.ucdavis.edu, to designate a particular computer. The domain name system is
an Internet service consisting of servers that translate host names into IP addresses.
Other Internet Protocols

The Internet Protocol Suite contains several protocols other than TCP and IP. These
include application layer protocols such as File Transfer Protocol (FTP) for transferring files between computers; Simple Mail Transfer Protocol for e-mail; TELNET,
which allows a person at a computer to use applications on a remote computer; and
Hypertext Transfer Protocol (HTTP), discussed below.
A universal resource locator (URL) is a string of characters used to obtain a service
from another computer on the Internet. Usually, the first part of a URL specifies the
protocol (e.g., FTP, TELNET, or HTTP), the second part of the URL is the host name
of the computer from which the resource is requested, and the third part specifies
the location of the file on the destination computer. For example, in the URL http://
www.ucdmc.ucdavis.edu/physics/text/, http is the protocol, www.ucdmc.ucdavis.
edu, is the host name, and physics/text/ identifies the location of the information
on the host computer.
The World Wide Web (WWW) is the largest and fastest growing use of the Internet. The WWW consists of servers connected to the Internet that store documents,
called web pages, written in a language called Hypertext Markup Language (HTML).
To read a web page, a person must have a computer, connected to the Internet, that
contains an application program called a web browser such as Firefox or Microsoft
Internet Explorer. Using the web browser, the user types or selects a URL designating
the web server and the desired web page. The browser sends a message to the server
requesting the web page and the server sends it, using a protocol called Hypertext
Transfer Protocol (HTTP). When the browser receives the web page, it is displayed
using the HTML instructions contained in the page. For example, the URL http://
www.ucdmc.ucdavis.edu/physics/text/ will obtain, from a server at the U.C. Davis
Medical Center, a web page describing this textbook.
Much of the utility of the WWW stems from the fact that a displayed web page
may itself contain URLs. If the user desires further information, he or she can select
a URL on the page. The URL may point to another web page on the same server or
it may point to a web page on a server on the other side of the world. In either case,
the web browser will send a message and receive and display the requested web page.
WWW technology is of particular interest in PACS because it can be used to provide
images and reports to clinicians.
142
Long Distance Telecommunications Links

Long distance telecommunication links are used to connect individual computers
to a distant network and to connect distant LANs into a WAN. Most long distance
telecommunication links are provided by telecommunications companies. In some
cases, there is a fixed fee for the link, depending upon the maximal bandwidth and
distance, regardless of the usage. In other cases, the user only pays for the portion of
time that the link is used. Connections to the Internet are usually provided by companies called Internet service providers (ISPs). The interface to an ISPs network can
be by nearly any of the telecommunications links described below.
The slowest but least expensive link is by the telephone modem. Modem is a
contraction of modulator/demodulator. A telephone modem converts digital information to a form for transmission over the standard voice-grade telephone system to
another modem. Typical data transfer rates of modems are up to 56 kbps.
Digital subscriber lines (DSL) use modulation techniques similar to a modem
to transfer digital information over local copper twisted-pair telephone lines to
the nearest telephone company end office at speeds up to about 1.5 Mbps. At
the telephone companys end office, the information is converted to another form
suitable for the telephone companys or an ISPs network. DSL is usually used to
connect computers or networks to the Internet. Some versions of DSL provide
higher speeds downstream (to the customer) than upstream and are referred to
as asymmetric DSL (ADSL). DSL is usually available only when the length of the
copper telephone lines from the home or office to the nearest telephone company
end office is less than about 5 km; the bandwidth decreases as the length of the
line increases.
Cable modems connect computers to the Internet by analog cable television lines.
Cable television lines usually support a higher bandwidth than do local telephone
lines, often up to or beyond 2 Mbps. A disadvantage to using cable television lines is
that the bandwidth is shared among the users of the line. If only a single computer
is connected to the line, it can utilize the entire bandwidth; however, if several
computers are connected, they share the bandwidth.
Some local telecommunication companies provide optical fiber cables directly
to homes and small businesses. These usually provide higher bandwidth than
DSL and cable modems. These lines will likely ultimately replace DSL and cable
modems.
Point-to-point digital telecommunications links may be leased from telecommunications companies. A fixed rate is usually charged, regardless of the amount of
usage. These links are available in various capacities (Table 5-9). The most common
are T1, at 1.544 Mbps, and T3, at 44.7 Mbps. Although these links behave as dedicated lines between two points, they are really links across the telecommunications
companys network. Optical carrier (OC) links are high-speed links that use optical
fiber transmission lines.
The Internet itself may be used to link geographically separated LANs into a
WAN. Encryption and authentication can be used to create a virtual private network (VPN) within the Internet. However, a disadvantage to using the Internet
to link LANs into a WAN is the inability of todays Internet to guarantee quality
of service. Disadvantages to the Internet today include lack of reliability, inability to guarantee bandwidth, and inability to give critical traffic priority over less
important traffic. For critical applications, such as PACS and teleradiology, quality
of service is the major reason why leased lines are commonly used instead of the
Internet to link distant sites.
143
TABLE 5-9LONG DISTANCE COMMUNICATIONS LINKS PROVIDED BY
TELECOMMUNICATION COMPANIES
DATA TRANSFER RATE
NOTE
Modem
28.8, 33.6, and 56 kbps
Digital signals over

No. Can connect to
commercial telephone system any computer or ISP
with modem.
POINT-TO-POINT?
DSL
Various, typically
up to DS1.
Digital signals over local

telephone wires
Typically provides
connection to one ISP.
Cable modem Various, typically

500 kbps to 10 Mbps
Signals over a cable

television line
Typically provides
connection to one ISP.
T1
DS1 (1.544 Mbps)
Leased line
Yes
T3
DS3 (44.7 Mbps)
Leased line
Yes
OC1
51.84 Mbps
Leased line
Yes
OC3
155 Mbps
Leased line
Yes
OC48
2,488 Mbps
Leased line
Yes
These can be used to connect an individual computer to a network or to link LANs into a WAN.
5.7 PACS and Teleradiology

Digital imaging technology and the falling costs and increasing capabilities of computer
networks, workstations, storage devices and media, and display devices have spurred
the implementation of picture archiving and communications systems (PACS) and
teleradiology, with the goal of improving the utilization and efficiency of radiological
imaging. For example, a PACS can replicate images at different display workstations
simultaneously for the radiologist and referring physician, and be a repository for
several years images. The loss of images, once a major problem in teaching hospitals producing images on film, has been mostly eliminated. The transfer of images
via teleradiology can bring subspecialty expertise to rural areas and smaller facilities
and permit prompt interpretation of studies, particularly in trauma and other urgent
cases, outside of normal working hours. Knowledge of the basic aspects of these systems is needed when obtaining or upgrading them; for their optimal use; and when
purchasing imaging devices that will exchange information with them.
A PACS is a system for the storage, transfer, and display of radiological images.
Teleradiology is the transmission of such images for viewing at a site or sites remote
from where they are acquired. PACS and teleradiology are not mutually exclusive.
Many PACS incorporate teleradiology. It is essential that PACS and teleradiology systems provide images suitable for the task of the viewer and, when the viewer is
the interpreting physician, the images viewed must not be significantly degraded in
either contrast or spatial resolution with regard to the acquired images.
Picture Archiving and Communications Systems

A PACS consists of a digital archive to store medical images, display workstations to
permit physicians to view the images, and a computer network to transfer images and
related information between the imaging devices and the archive and between the
archive and the display workstations. There also must be a database program to track
144
the locations of images and related information in the archive and software to permit
interpreting, and perhaps referring, physicians to select and manipulate images. For
efficiency of workflow and avoidance of errors, the PACS should exchange information
with other information systems, such as the hospital information system (HIS), radiology information system (RIS), and electronic medical record (EMR) system. There
may be additional components to the PACS, such as a film digitizer, a film printer, a
CD or DVD burner, and a web server to provide images to referring clinicians.
Today, PACS vary widely in size and in scope. On one hand, a PACS may be
devoted to only a single modality at a medical facility. For example, a PACS may be
limited to a nuclear medicine department, the ultrasound section, mammography,
or a cardiac catheterization laboratory. Such a small single-modality PACS is sometimes called a mini-PACS. Mini-PACSs may exist in a large medical enterprise that has
adopted an enterprise-wide PACS, if the enterprise PACS lacks functionality needed
by specialists such as mammographers or nuclear medicine physicians. On the other
hand, a PACS may incorporate all imaging modalities in a system of several medical
centers and affiliated clinics (Fig. 5-23). Furthermore, a PACS may permit images to
be viewed only by interpreting physicians or it may make them available to the ER,
ICUs, and referring physicians as well. The goal, achieved at many medical centers
today, is to store all images in a medical center or healthcare system on PACS, with
images available to interpreting and referring clinicians through the EMR or thin-
client workstations within the enterprise, with the PACS receiving requests for studies from the HIS and RIS, and with the PACS providing information to the HIS and
RIS on the status of studies, and images available on the EMR. Another goal, far from
being achieved, is to make medical images and related reports available nationwide,
regardless of where they were acquired.
Imaging
Modalities
Exam orders
Scheduling
Reporting
HL7
DICOM
MWL
RIS
EMR
PACS Wide Area

Archive Network
VPN
Off Site
Backup
Internet
VPN
Radiology Reading
Reporting
Web
Server
Emergency
Backup Enterprise Image
CustomerCare
Distribution
Monitoring
Server
FIGURE 5-23 Modern PACS uses a webserver connected to the Internet and a thin client paradigm to
allow referring clinicians without access to the medical enterprise network to obtain images and reports over
the Internet. Users within the medical enterprise have access through a LAN and WAN. The RIS provides the
patient database for scheduling and reporting of image examinations through HL7 (Health Level7) transactions and provides modality worklists (MWLs) with patient demographic information to the modalities, allowing technologists to select patient-specific scheduled studies to ensure accuracy. After a study is performed,
the information is sent to the PACS in DICOM format, and reconciled with the exam-specific information
(accession number). An emergency backup server ensures business continuity (red dotted line and highlights)
in the event of unscheduled PACS downtime. Also depicted are an off-site backup archive for disaster
recovery and real-time customer care monitoring to provide around the clock support. Not shown is the mirror
archive, which provides on-site backup, with immediate availability, in cause of failure of the primary archive.
145
Teleradiology
Teleradiology can provide improved access to radiology for small medical facilities
and improved access to specialty radiologists. For example, teleradiology can permit radiologists at a medical center to promptly interpret radiographs acquired at
affiliated outpatient clinics. This requires devices producing digital radiographs or
film digitizers at each clinic, leased lines to the medical center, and an interpretation workstation at the medical center. In another example, teleradiology can permit
an on-call radiologist to interpret after-hours studies at home, avoiding journeys to
the medical center and providing interpretations more promptly. This requires the
radiologist to have an interpretation workstation at home, perhaps with a DSL or
cable modem connection provided by a local telecommunications company, and the
medical center to have images in a digital format and a matching communications
connection. Image interpretation outside normal working hours may be provided
by commercial nighthawk services consisting of radiologists with display workstations, sometimes located in countries such as Australia or India with time zones very
different from those in the Americas, connected by high bandwidth links provided
by telecommunications companies.
Acquisition of Digital Images

Most medical imaging devices today produce digital images. However, radiographic
and mammographic images at some medical centers are still acquired using
film-screen image receptors and older images may exist on film. Film images can be
digitized by laser or charge-coupled device (CCD) digitizers.
Image Formats
The formats of digital radiological images are selected to preserve the clinical information acquired by the imaging devices. As mentioned earlier in this chapter, the
numbers of rows and columns of pixels in an image are determined by the spatial resolution and the field-of-view of the imaging device, whereas the number of bits used
for each pixel is determined by the contrast resolution of the imaging device. Thus,
imaging modalities providing high spatial resolution (e.g., mammography) and with
large fields of view (e.g., radiography) require large numbers of pixels per image,
whereas modalities providing lower spatial resolution (e.g., CT) or small fields of
view (e.g., ultrasound) can use fewer pixels. Modalities providing high contrast resolution (e.g., x-ray CT) require a large number of bits per pixel, whereas modalities
with low contrast resolution (e.g., ultrasound) require fewer bits per pixel. Typical
formats of digital images are listed in Table 5-7.
Film Digitizers
A film digitizer is a device that uses a light source and one or more light detectors to
measure the light transmitted through individual areas of a film and form a digital
image depicting the distribution of optical density in the film. The most common
digitizer technology employs a collimated light source and a charge-coupled-device
(CCD) linear array.
In a CCD digitizer (Fig. 5-24), the film is continuously moved across a lighted slit.
A lens system focuses the transmitted light onto a CCD. A CCD, discussed in Chapter
7, is a solid-state integrated circuit with a rectangular array of light detectors. Each
detector element of the CCD accumulates the signal from the transmitted light incident
on it and stores the signal as an electrical charge until the CCD is read. When the CCD
146

Film lnput
Digital lmage
Assemly
LUT
Conversion
ADC & Amplification
Digital Image
Processing
Linear CCD with

micro-lens Array
Collimated
Light Source
DICOM
Information
Image to
PACS
Transport
Rollers
Film Output
FIGURE 5-24 CCD film digitizer. A linear array of light-emitting diodes is the typical illumination source and
a CCD array at near contact to the film detects the transmitted light, which is then converted into a corresponding digital value, typically using 16 bits of quantization for modern CCD digitizers. In this configuration,
the linear CCD photo-sensor, comprised of up to 4,000 pixels in a single row, produces electrons in proportion
to the light transmitted through the film at a corresponding film location. The signal is proportional to the
fraction of transmitted light.
is read, the analog signal from each element of the CCD is digitized by an analogto-digital converter (ADC). The logarithm is taken of the intensity signal from each
element of the CCD so that each pixel value in the digital image is proportional to the
optical density (OD) of the corresponding portion of the film. (In Chapter 7, the optical
density is defined as OD 5 log10 (Io/I) 5 log10 (Io) 2 log10 (I), where I is the intensity of
the light transmitted through the film and Io is the intensity if the film is removed.)
The role of film digitizers in PACS is likely to be transient, due to the decreasing use of screen-film image receptors. However, it is likely be several years before
the films role in recording and displaying images will be but a footnote in a future
edition of this book.
Transfers of Images and Related Information

Computer networks, discussed in detail earlier in this chapter, permit exchanges of
images and related information between the imaging devices and the PACS, between
the PACS and display workstations, between the PACS and other information systems
such as the RIS and HIS, and between various components of the PACS itself. A PACS
may have its own local area network (LAN) or LAN segment, or it may share another
LAN, such as a medical center LAN. The bandwidth requirements depend upon the
imaging modalities and their workloads. For example, a LAN adequate for nuclear
medicine or ultrasound may not be adequate for other imaging modalities. Network
traffic typically varies in a cyclical fashion throughout the day. Network traffic also
tends to be bursty; there may be short periods of very high traffic, separated by periods of low traffic. Network design must take into account both peak and average bandwidth requirements and the delays that are tolerable. Network segmentation, whereby
groups of imaging, archival, and display devices that communicate frequently with
each other are placed on separate segments, is commonly used to reduce network congestion. Network media providing different bandwidths may be used for various net-
147
work segments. For example, a network segment serving nuclear medicine will likely
have a lower bandwidth requirement than a network segment serving CT scanners.
Digital Imaging and Communications in Medicine

Connecting imaging devices to a PACS with a network, by itself, does not achieve
transfer of images and related information. This would permit the transfer of files,
but medical imaging equipment manufacturers and PACS vendors may use proprietary formats for digital images and related information. In the past, some facilities
solved this problem by purchasing all equipment from a single vendor; others had
custom software developed to translate one vendors format into anothers format.
To help overcome problems such as these, the American College of Radiology and
the National Electrical Manufacturers Association jointly sponsor a set of standards
called Digital Imaging and Communications in Medicine (DICOM) to facilitate the transfer of medical images and related information. Other professional societies work to
develop medical specialty-specific DICOM standards. Many other national and international standards organizations recognize the DICOM standards.
DICOM includes standards for the transfer, using computer networks, of images
and related information from individual patient studies between devices such as imaging devices and storage devices. DICOM specifies standard formats for the images
and other information being transferred, services that one device can request from
another, and messages between the devices. DICOM does not specify formats for
the storage of information by a device itself, although a manufacturer may choose to
use DICOM formats for this purpose. DICOM also includes standards for exchanging information regarding workflow; standards for the storage of images and related
information on removable storage media, such as optical disks; and standards for the
consistency and presentation of displayed images.
DICOM specifies standard formats for information objects, such as patients,
images, and studies. These are combined into composite information objects,
such as the DICOM CT (computed tomography) image object, CR (computed radiography) image object, DX (digital x-ray) image object, MG (digital mammography x-ray) image object, US (ultrasound) image object, MR (magnetic resonance
imaging) image object, and NM (nuclear medicine) image object. DICOM specifies
standard services that may be performed regarding these information objects, such
as storage, query and retrieve, storage commitment, print management, and media
storage.
DICOM provides standards for workflow management, such as modality worklists
(MWLs) listing patients to be imaged on specific imaging devices, and Performed
Procedure Step, for communicating information about the status of a procedure.
DICOM also provides standards, particularly Grayscale Standard Display Function
(GSDF) and Presentation State, for the consistency and presentation of displayed and
printed images. The DICOM GSDF was discussed earlier in this chapter.
DICOM Presentation State is a standard for capturing and storing a technologists
or a radiologists adjustments of and annotations on key images. It records adjustments such as roaming and zooming, cropping, flipping, and windowing and leveling, and annotations such as arrows and clinical notes. These are stored with the
image set on a PACS. When the images are viewed, the key images appear with the
adjustments and annotations made by the technologist or radiologist.
DICOM is compatible with common computer network protocols. DICOM is
an upper layer (Application Layer) standard and so is completely independent of
lower layer network protocols, such as LAN protocols. DICOM adopts the Internet
148
Protocol Suite (TCP/IP). DICOM will function on any LAN or WAN, provided that
the intermediate network layer protocol is TCP/IP.
Today, the products of nearly all vendors of medical imaging and PACS equipment permit exchanges of information that conform to parts of the DICOM standard.
A DICOM conformance statement is a formal statement, provided by a vendor, describing a specific implementation of the DICOM standard. It specifies the services, information objects, and communications protocols supported by the implementation.
There are practical issues regarding DICOM that are worthy of emphasis. First,
DICOM applies not just to a PACS, but also to each imaging device that exchanges
information with the PACS. Hence, issues of DICOM conformance and functionality
must be considered when purchasing individual imaging devices, as well as when
purchasing or upgrading a PACS. Another issue is that DICOM is a set of standards,
not a single standard. When purchasing an imaging device or a PACS, the contract
should specify the specific DICOM standards with which conformance is desired. For
example, support for DICOM Modality Worklists, Performed Procedure Step, and
Storage Commitment should be provided in most cases, in addition to image store.
Furthermore, there may be more than one DICOM standard that will permit information transfer, but all may not be equally useful. For example, digital radiographs may
be transferred as the older DICOM CR image object or the newer DX image object.
However, the old CR image object contains much less mandatory information regarding procedure, projection, laterality, etc. Also, the CR image object does not clearly
define the meaning of pixel values. Even though use of the CR image object will
permit image transfer, the lack of information about the image will hinder the use
of automatic hanging protocols and optimal display of image contrast. For example,
purchasers of PACS and CR and DR imaging systems should consider whether the
systems support transfers using the CR image object or the newer DX image object
and whether display workstations are able to use the additional information in the
DX image object to provide automated hanging protocols (Clunie, 2003).
Communication with the Radiology Information System

and Hospital Information System
It is advantageous to have communication among a PACS, the radiology information system (RIS), the hospital information system (HIS), and the electronic medical record (EMR, also known as the electronic patient record and electronic health
record). A RIS is an information system that supports functions within a radiology
department such as ordering and scheduling procedures, maintaining a patient database, transcription, reporting, and bill preparation. The RIS is not always a separate
systemit may be part of the HIS or incorporated in the PACS. The RIS can provide worklists of scheduled studies to the operators consoles of the imaging devices,
thereby reducing the amount of manual data entry and the likelihood of improperly
identified studies.
Communication among the RIS, HIS, EMR, and PACS is often implemented using
a standard called HL7 (Health Level 7) for the electronic exchange of alphanumeric
medical information, such as administrative information and clinical laboratory
data. HL7 to DICOM translation, for instance, is a step in the RIS or HIS providing
worklists to the imaging devices. HL7, like DICOM, applies to the Application Layer
in the network protocol stack.
Modality worklists, supplied by the RIS, contain patient identifying information
that helps to reduce a common problem in PACSthe inadvertent assignment of
different patient identifiers to studies of a specific patient. This can arise from errors
149
in manual data entry, for example, a technologist or clerk entering a single digit of a
social security number incorrectly. Entering names differently (e.g., Samuel L. Jones,
S. L. Jones, and Jones, Samuel) also may cause problems. These problems are largely
obviated if the imaging technologists select patient identifiers from worklists, provided of course that the RIS and/or HIS furnishes unique patient identifiers. A lost
study can still occur if the technologist mistakenly associates an imaging study with
the wrong patient in a worklist.
Modality worklists also permit the prefetching of relevant previous studies for
comparison by interpreting physicians. The network interfaces to the RIS and HIS
can provide the interpreting physicians with interpretations of previous studies and
the electronic patient record via the PACS workstation, saving the time that would
be required to obtain this information on separate computer terminals directly connected to the RIS and/or HIS. These interfaces can also permit the PACS to send information to the RIS and/or HIS regarding the status of studies (e.g., study completed).
In a PACS, each patient study is commonly identified by a unique number called
the accession number. The accession number is commonly assigned by the RIS.
Integrating the Healthcare Enterprise

As mentioned above, to provide optimal efficiency, the PACS and imaging devices
must exchange information with other healthcare information systems. DICOM and
HL7 provide standards for such exchanges of information, but they do not fully
describe the optimal use of these standards within a healthcare system. IHE (Integrating the Healthcare Enterprise) is an initiative sponsored by the Radiological Society of
North America and the Health Information Management Systems Society to improve
the sharing of information by healthcare information systems. Under IHE, priorities
for integration are identified and industry representatives achieve consensus on
standards-based transactions to meet each identified clinical need. These decisions
are recorded in documents called integration profiles, the collection of which is
called the Technical Framework and made freely available. Participating companies
build into their systems the capability to support IHE transactions. Testing, performed during events called Connectathons, ensure a high degree of conformity with
the Technical Framework. IHE does not provide standard formats for the exchange
of information between pairs of devices. It instead describes the uses of existing
standards such as DICOM, HL7, and the Internet Protocol Suite.
The most fundamental IHE integration profile for radiology is Radiology Scheduled Workflow. It defines the flow of information for the key steps in a typical patient
imaging encounter (registration, ordering, scheduling, acquisition, distribution, and
storage). If an imaging device or PACS conforms to the Scheduled Workflow Integration Profile, it will support DICOM store, modality worklist, performed procedure
step, and storage commitment transactions. At the time of publication of this textbook, the IHE Technical Framework included about 20 integration profiles relevant
to radiology that provide solutions to actual clinical problems. For example, significant hindrances to workflow can occur in mammography if the digital mammography system, the PACS or mammography mini-PACS, and interpretation workstation
do not conform to the IHE Mammography Image Integration Profile in addition to
the Radiology Scheduled Workflow Integration Profile. A major benefit of the IHE
integration profiles is that a single integration profile can require conformance with
several DICOM, HL7, and other standards and so a commitment by a vendor to support that integration profile will commit the vendor to conforming with the various
standards included in that single integration profile.
150
Provision of Images and Reports to Attending Physicians

and Other Healthcare Providers
WWW technology, described above, provides a cost-effective means for the distribution of images and other information over the HIS to noninterpreting physicians and
other healthcare professionals. A web server, interfaced to the PACS, RIS, and HIS,
maintains information about patient demographics, reports, and images. The server
permits queries and retrievals of information from the PACS database. Physicians
obtain images and reports from the web server using workstations with commercial
web browsers. Information is sent from the PACS, typically in a study summary form
containing just the images that are pertinent to a diagnosis. The full image set may be
available as well. A major advantage to using WWW technology is that the workstations need not be equipped with specialized software for image display and manipulation; instead, these programs can be sent from the web server with the images or
the server itself can provide this function. It is particularly desirable, for healthcare
systems that have implemented electronic medical records, that relevant images can
be accessed from the EMR.
Healthcare professionals at remote sites such as doctors offices, small clinics,
or at home can connect to the EMR using modems or leased lines. Alternatively, if
the PACS web server is interfaced to the Internet, these clinicians can obtain images
and reports over the Internet using DSL lines, cable modems, or fiber optic links
provided by Internet service providers.
Storage of Images
The amount of storage required by a PACS archive depends upon the modality or
modalities served by the PACS, on the workload, and the storage duration. The
amounts of storage required for individual images and for typical studies from the
various imaging modalities were discussed earlier in this chapter. For example, a
medium size nuclear medicine department without PET/CT or SPECT/CT generates
a few gigabytes (uncompressed) of image data in a year and so a single large-capacity
magnetic disk and an optical disk drive for backup would suffice for a mini-PACS
serving such a department. On the other hand, CT, MRI, and digitized radiographs
from a medium size radiology department can generate several gigabytes of image
data in a day and several terabytes (uncompressed) in a year, requiring a much larger
and complex archival system. Ultrasound storage requirements strongly depend on
the number of images in a study that are selected for archiving. For a medium-sized
medical center, if 60 images per patient are selected for storage, then up to 0.5 TB
per year may be generated. A cardiac catheterization laboratory may generate a few
terabytes (uncompressed) in a year.
Storage Schemes
In a PACS, one of two storage schemes is commonly used. In a hierarchical storage
scheme, recent images are stored on arrays of magnetic hard disk drives, and older
images are stored on slower but more capacious archival storage media, such as
optical disks and magnetic tape. In hierarchical storage schemes, the term on-line
storage is used to describe storage, typically arrays of magnetic disks, used to provide
nearly immediate access to studies. The term near-line storage refers to storage, such
as automated libraries of optical disks or magnetic tape, from which studies may be
retrieved within about a minute without human intervention. Optical disks have
an advantage over magnetic tape cassettes for near-line storage because of shorter
151
a verage access times, but magnetic tape is less expensive per unit storage capacity.
Off-line storage refers to optical disks or magnetic tape cassettes stored on shelves or
racks. Obtaining a study from off-line storage requires a person to locate the relevant
disk, cartridge, or cassette and to manually load it into a drive. When hierarchical
storage is used, if considerable delays in displaying older studies are to be avoided,
the system must automatically copy (prefetch) relevant older studies from near-line
to online storage to be available for comparison when new studies are viewed.
An alternative to hierarchical storage is to store all images on arrays of magnetic
hard disk drives. As these become full, more disk arrays are added to the system. This
storage scheme has become feasible because of the increasing capacities of magnetic
drives and the decreasing cost per unit storage capacity. This method is commonly
referred to as everything on line storage.
Regardless of the storage scheme, the PACS must create and maintain a copy or
copies of all information on separate storage devices for backup and disaster recovery. This is discussed later in this chapter in the section on security. The backup storage should be in a location such that a single plausible incident would not destroy
both the primary and backup information.
Image Management
The PACS archive may be centralized, or it may be distributed, that is, stored at
several locations on a network. In either case, there must be archive management
software on a server. The archive management software includes a database program
that contains information about the stored studies and their locations in the archive.
The archive management software communicates over the network with imaging
devices sending studies for storage and sends copies of the studies received from
imaging devices to the storage devices, including to backup storage. The transfers
between imaging devices and the PACS usually conform to the DICOM standard.
The archive management software must also obtain studies from the storage devices
and send either the studies or selected images from them to workstations requesting
studies or images for display. In PACS with hierarchical storage, the archive management software transfers studies between the various levels of archival storage, based
upon factors such as the recentness of the study and, when a new study is ordered,
prefetches relevant older studies from near-line storage to online storage to reduce
the time required for display.
In some PACS, studies or images from studies awaiting interpretation and relevant
older studies are requested by viewing workstations from the PACS archive as needed
(on demand) during viewing sessions, but this can slow the workflow process of
the interpreting physician. Alternatively, studies may be obtained (prefetched) from
the archive and stored on a display workstation or local file server, prior to the interpretation session, ready for the interpreting physicians use. The prefetch method
requires the interpretation workstations or server to have more local storage capacity,
whereas the on-demand method requires a faster archive and faster network connections between the archive and the interpretation workstations. An advantage to the
on-demand method is that a physician may use any available workstation to view a
particular study, whereas to view a particular study with the prefetch method, the
physician must go to a workstation that has access to the locally stored study. When
images are fetched on-demand, the first image should be available for viewing within
about two seconds. Once images reside on the local workstations disk or local server,
they are nearly instantaneously available.
On-demand systems may send all the images in entire studies to the workstation,
or they may just send individual images when requested by the viewing workstation.
152
The method of sending entire studies at once requires greater network bandwidth
and much more magnetic disk storage capacity on the viewing workstation and
causes greater delay before the first images are displayed, but once the images are
stored on the viewing workstations disk, they are nearly instantaneously available.
On the other hand, providing only individual images upon request by the viewing workstation reduces the delay before the first image or images are displayed,
but places more demand on the archive server to respond to frequent requests for
individual images.
Image Compression
The massive amount of data in radiological studies (Table 5-10) poses considerable challenges regarding storage and transmission. Image compression reduces the
number of bytes in an image or set of images, thereby decreasing the time required
to transfer images and increasing the number of images that can be stored on a
magnetic disk or unit of removable storage media. Compression can reduce costs by
permitting the use of network links of lower bandwidth and by reducing the amount
of required storage. Before display, the image must be decompressed (Fig. 5-25).
Image compression and decompression can be performed by either a generalpurpose computer or specialized hardware. Image compression and decompression
are calculation-intensive tasks and can delay image display. The compression ratio is
the number of bytes of the uncompressed image divided by the number of bytes of
the compressed image.
There are two categories of compression: reversible, also called bit-preserving,
lossless, or recoverable compression; and irreversible, also called lossy or nonrecoverable, compression. In reversible compression, the uncompressed image is identical to the original. Typically, reversible compression of medical images provides
compression ratios from about two to three. Reversible compression takes advantage of redundancies in data. It is not possible to store random and equally likely
bit patterns in less space without the loss of information. However, medical images
incorporate considerable redundancies, permitting them to be converted into a more
compact representation without loss of information. For example, although an image
may have a dynamic range (difference between maximal and minimal pixel values)
requiring 12 bits per pixel, pixel values usually change only slightly from pixel to
adjacent pixel and so changes from one pixel to the next can be represented by just
TABLE 5-10IMAGE STORAGE REQUIREMENTS FOR VARIOUS
RADIOLOGICAL STUDIES, IN UNCOMPRESSED FORM

STUDY
TYPICAL STORAGE (MB)
Chest radiographs (PA and lateral, 2 3 2.5 k)
20
Standard CT series of the head (50 512 images)
26
Tl-201 myocardial perfusion SPECT study

Ultrasound (512 , 60 images to PACS archive)
2
Cardiac cath lab study (coronary and LV images)

Digital screening mammograms (2 CC and 2 MLO)
1
16
4503,000
60132
Note: The number of images per study varies greatly depending on the type of study (e.g., CT cardiac
angiography), number of acquisition phases, and whether thin slices for MPRs are necessary, which can
increase the number of images by an order of magnitude. Storage is calculated as the product of the number
of images and the size in bytes of an image.
153
FIGURE 5-25 Image compression reduces the number of bytes in an image to reduce image storage space
and image transmission times. At a display workstation, an image retrieved from the archive over the network
require image decompression, which restores the images to full physical size (number of pixels) and number of
bytes. Shown above is a chest image with lossless compression (left), and 30:1 jpeg lossy compression (right).
Although the minified images (above) look similar, the magnified views (below) illustrate loss of image fidelity
and nondiagnostic image quality with too much lossy compression.
a few bits. In this case, the image could be compressed without a loss of information
by storing the differences between adjacent pixel values instead of the pixel values
themselves. Dynamic image sequences, because of similarities from one image to the
next, permit high compression ratios.
In irreversible compression, information is lost and so the uncompressed image
will not exactly match the original image. However, irreversible compression permits much higher compression; ratios of 15-to-one or higher are possible with very
little loss of image quality. Currently, there is controversy on how much compression can be tolerated. Research shows that the amount of compression is strongly
dependent upon the type of examination, the compression algorithm used, and
the way that the image is displayed (e.g., on film or display monitor). In some
cases, images that are irreversibly compressed and subsequently decompressed are
actually preferred by radiologists over the original images, due to the reduction of
image noise. Legal considerations also affect decisions on the use of irreversible
compression in medical imaging. The ACR Technical Standard for Electronic Practice
of Medical Imaging (2008) states, Data compression may be performed to facilitate transmission and storage. The type of medical image, the modality, and the
objective of the study will determine the degree of acceptable compression. Several methods, including both reversible and irreversible techniques may be used
under the direction of a qualified physician or practitioner, with minimal if any
reduction in clinical diagnostic image quality. If compression is used, algorithms
recommended by the DICOM standard should be used. The types and ratios
154
of compression used for different imaging studies transmitted and stored by the
system should be selected and periodically reviewed by the responsible physician to
ensure appropriate clinical image quality.
However, at the time of printing, the US Food and Drug Administration (FDA),
under the authority of the federal Mammography Quality Standards Act, restricts
the use of lossy compression in mammography. In particular, the FDA does not
permit mammograms compressed by lossy methods to be used for final interpretation, nor does it accept the storage of mammograms compressed by lossy methods
to meet the requirements for retention of original mammograms. The FDA does
permit interpretation of digital mammograms compressed by lossless methods and
considers storage of such mammograms to meet the requirement for retention of
original mammograms. The reader should refer to current guidance from the FDA
on this topic.
Display of Images
Images from a PACS may be displayed on monitors or, less often, may be recorded
by a laser imager on light-sensitive film, which is chemically developed, and then
viewed on light boxes. The purpose of the display system is to present anatomic
and/or physiologic information in the form of images to interpreting physicians and
other clinicians.
Computer workstations equipped with display monitors are commonly used,
instead of viewboxes and photographic film, to display radiological images for
both interpretation and review. The various radiological modalities impose specific
requirements on workstations. For example, a workstation that is suitable for the
interpretation of nuclear medicine or ultrasound images may not be adequate for
digital radiographs or mammograms.
As mentioned earlier in this chapter, the CRT monitors once used for radiological
image display have been mostly replaced by high-quality LCD monitors. The physical principles and performance of these devices were described earlier in this chapter
and only their use in medical imaging will be discussed here.
Display Workstations
Display Systems for Workstations
As mentioned above, the various imaging modalities impose specific requirements on
workstations. An interpretation workstation for large matrix images (digital radiographs,
including mammograms) is commonly equipped with two high-luminance 54-cm
diagonal 3 or 5 megapixel grayscale monitors, in the portrait orientation, to permit the
simultaneous comparison of two images in near full spatial resolution (Fig.5-26). A
navigation monitor is also present on a diagnostic workstation to provide access to
the RIS database, patient information, reading worklist, digital voice dictation system,
EMR, and the Internet. Such a workstation is also suitable for viewing CT and MRI
images, unless surface and volume rendering are utilized, in which case color monitors
are preferable. Such a workstation can also be used to view digital fluorographs and, if
it has adequate cine capability, angiographic images. Three megapixel 54-cm diagonal
color monitors are available with sufficient calibrated maximal luminances (400 to 500
cd/m2) for viewing radiographs, and are now a typical choice for replacing older grayscale monitors. Another monitor format that is becoming popular is the 76-cm diagonal monitor, available with 4, 6, or 10 megapixels, designed to replace a pair of 54-cm
155
FIGURE 5-26 Interpretation workstation containing two 1.5 by 2 k pixel (3 megapixel) portrait-format
grayscale monitors for high resolution and high luminance image interpretation, flanked by two 1.9 by 1 k
(2megapixel MP) color navigation monitors (left and right) for PACS access, patient worklist, timeline, and
thumbnail image displays; multipurpose display for 3D color rendering, digital voice dictation reporting, and
EMR and RIS information access. The keyboard, mouse, and image navigation and voice dictation device assist
Ramit Lamba, M.D., in his interpretation duties.
diagonal monitors. Such a monitor provides a seamless viewing area. However, when
a pair of 54-cm monitors is used, each can be turned slightly inward to face the viewer,
whereas, when a 76-cm monitor displays a pair of radiographs, the viewer must move
his or her head from side to side to look directly at each radiograph.
Workstations for viewing CT and MRI images, fluorographs, and angiographic
images can have smaller pixel format monitors. If surface or volume rendered images
or co-registered PET/CT or SPECT/CT images are to be displayed, the workstation
should be equipped with high-luminance color monitors. Workstations for interpretation of angiographic image sequences (e.g., cardiac angiograms and ventriculograms and digital subtraction angiograms) should be capable of displaying cine
images (sequences of images acquired to depict temporal variations in anatomy or
contrast material).
Workstations for the interpretation of nuclear medicine and ultrasound images
should have color monitors and must be able to display cine images. A single color
monitor with a pixel format on the order of 1,000 3 1,000 is sufficient for a nuclear
medicine or ultrasound workstation.
Most noninterpreting physicians typically use standard commercial personal
computers with consumer-grade color monitors, although some, such as orthopedic
surgeons, may require specialized workstations with display systems similar to those
of interpreting physicians.
An application program on the workstation permits the interpreting physician to
select images, arrange them for display, and manipulate them. The way in which the
program arranges the images for presentation and display is called a hanging protocol.
The hanging protocols should be configurable to the preferences of individual physicians. Image manipulation capabilities of the program should include window and
level, and zoom (magnify) and roam (pan). In particular, because the dynamic range
of a display monitor is much less than that of film on a viewbox, windowing and
leveling are needed to compensate. The program should also provide quantitative
tools to permit accurate measurements of distance and the display of individual pixel
156
values in modality-appropriate units (e.g., CT numbers). Image processing, such as

smoothing and edge enhancement, may be provided. Modality-specific or physician specialtyspecific tools may be needed. For example, ultrasound and nuclear
medicine have specific display and tool requirements and orthopedic surgeons and
cardiologists have specific requirements such as specific quantitative and angular
measurements.
Ambient Viewing Conditions

Proper ambient viewing conditions, where interpretation workstations are used, are
needed to permit the interpreting physicians eyes to adapt to the low luminances of
the monitors, to avoid a loss in contrast from diffuse reflections from the faces of the
monitors, and to avoid specular reflections on the monitors faces from bright objects.
Viewing conditions are more important when monitors are used than when viewboxes are used, because of the lower luminances of the monitors. The faces of LCDs
are flat and less reflective than those of CRTs and so LCDs are less affected by diffuse
and specular reflections. The replacement of CRT monitors by LCD monitors has
made ambient viewing conditions less critical. The photometric quantity illuminance,
also discussed in Chapter 8, describes the rate of light energy, adjusted for the photopic spectral sensitivity of the human visual system, impinging on a surface, per unit
area. In areas used for viewing clinical images from display monitors, the illuminance
should be low (2040 lux), but not so low as to interfere with other necessary tasks,
such as reading printed documents, or to require a major adjustment in adaption
when looking at a monitor after looking at another object, or vice versa. A reading
room should have adjustable indirect lighting and should not have windows unless
they can be blocked with opaque curtains or shutters. When more than one workstation is in a room, provisions should be made, either by monitor placement or by the
use of partitions, to prevent monitors from casting reflections on each other. Particular care must be taken, if both conventional viewboxes and viewing workstations are
used in the same room, to prevent the viewboxes from casting reflections on monitor
screens and impairing the adaptation of physicians eyes to the dimmer monitors.
Challenges of PACS
There are many challenges to implementing, using, maintaining, upgrading, and
replacing PACS. An important goal is achieving efficient workflow. There are far too
many issues regarding efficiency of workflow to be listed here; this paragraph merely
lists several important objectives. Prior to image acquisition by an imaging device,
little manual entry of information should be required by imaging technologists.
This can be facilitated by utilizing modality worklists generated by the radiology
information system and sent to the modality. Worklists of studies to be interpreted,
arranged by factors such as modality and priority, should be provided to interpreting physicians. Images should be displayed promptly, ideally within two seconds,
to referring and interpreting physicians. Little or no manipulation of images should
be routinely necessary; automated hanging protocols should display the images in
the order, orientation, magnification, and position on the monitor expected. Software tools should be available to technologists to help identify studies that are not
assigned to the proper patient, are mislabeled, or require image manipulation. Image
interpretation is made more efficient by a digital speech recognition system that
promptly displays the dictated report for review and approval by the interpreting
physician before he or she interprets the next study. These systems greatly decrease
157
the turn-around-time of diagnostic reports, and are a major benefit to the referring
physician and ultimately to the patient.
Major challenges include the following:
1. Initial and recurring equipment purchase or lease costs.
2. Expensive technical personnel to support the PACS.
3. Training new users in system operation.
4. Achieving communication among equipment from several vendors.
5. Achieving information transfer among digital information systems (PACS, RIS,
HIS, EMR).
6. Maintaining security of images and related information and compliance with
Health Insurance Portability and Accountability Act (HIPAA) Regulations (discussed in a later section).
7. Equipment failures can impede interpretation and distribution of images and
reports.
8. Digital storage media longevity and obsolescence.
9. Maintaining access to images and data during PACS upgrades.
10. Maintaining access to archived images and related data if: converting to a new
archival technology, the PACS vendor goes out of business, the PACS vendor
ceases to support particular the PACS, or you replace the system with a PACS
from another vendor.
11. Providing access to and display of legacy film images.
12. Conversion of film images into digital images. (Some facilities choose to not
convert old film images to digital images.)
13. Completely eliminating the production of film images.
14. Locating misidentified studies.
15. Multiple PACS in a single healthcare enterprise.
16. Obtaining and importing images from outside your healthcare enterprise.
17. Providing images and reports to healthcare providers outside your healthcare
enterprise.
18. Quality control program for display monitors.
19. Providing adequate software tools for specialists such as orthopedic surgeons
and cardiologists.
This list is far from complete. Of course, the desirable strategy is to enhance the
advantages and minimize the disadvantages. Benefits will expand as experience with
these systems increases. Furthermore, the rapidly increasing capabilities and falling
costs of the technology used in PACS will continue to increase their performance
while reducing their cost.
Quality Control of PACS and Teleradiology Systems

PACS and teleradiology systems are part of the imaging chain and therefore require
quality control monitoring. Although both analog and digital portions of the system can degrade images, the performance of digital portions (computer network,
software, storage, etc.) regarding image quality does not usually change with time.
The performance of analog portions (film digitizers, laser printers, and display monitors) changes with time and therefore requires monitoring. Quality control can be
divided into acceptance testing, to determine if newly installed components meet
desired performance specifications, and routine performance monitoring. Display
monitors must be evaluated periodically to ensure all the monitors of a workstation
have adequate maximal calibrated luminances and the same display characteristics.
158
FIGURE 5-27 Society of Motion Picture and Television Engineers (SMPTE) monitor test pattern. This pattern
allows the inspection of spatial resolution in the center and at the corners of the image (high and low contrast
line patterns in horizontal and vertical orientations), assessment of contrast differences over the entire range
of luminance by the ten 10% intensity steps, contrast in the dark and bright areas of the image (5% patches
in the 0% and 100% bright areas, when used with a linear LUT), brightness linearity across the range of grayscale values, geometric linearity grid pattern, and transitions from dark to bright and bright to dark.
In particular, the display functions of grayscale monitors and color monitors that display grayscale images should be evaluated to ensure that they conform to the DICOM
GSDF. The maximal luminance of a display monitor degrades with the amount of
time that the monitor is on and with the brightness of displayed images. Turning
LCD monitors off when they will not be used for an extended period of time can
significantly prolong their lives; however, monitors may take some time to stabilize
from a cold start, so a common strategy is to place them into a standby mode.
Monitors should be replaced when their maximal luminances become inadequate.
Some medical grade LCD monitors can measure their own luminances and can send
information on their maximal calibrated luminances and conformance to the DICOM
GSDF across a network to a workstation, greatly simplifying the task of monitoring
display monitor performance.
Test patterns, such as the Society of Motion Picture and Television Engineers
(SMPTE) Test Pattern (Fig. 5-27) are useful for semi-quantitative assessments of display monitor performance. The American Association of Physicists in Medicine has
available a set of test images for monitor quality control and evaluation.** Monitor
faces should be cleaned frequently to remove fingerprints and other dirt to avoid
image artifacts.
Another important quality assurance task is ensuring that acquired studies are
properly associated with the correct patients in the PACS, RIS, and EMR. Problems
can occur if an error is made in manually entering patient identifying information, if a
technologist selects the wrong patient identifier from a modality worklist, an imaging
**http://aapm.org/pubs/reports/OR_03.pdf
159
study is performed on an unidentified patient, or if a patients name is changed, such

as after a marriage. These can result in imaging studies stored in a PACS that are not
associated with the proper patients. Procedures should be established to identify
such incorrectly associated studies and to associate them with the proper patients.
5.8 Image Processing

An important use of computers in medical imaging is to process digital images to
display the information contained in the images in more useful forms. Digital image
processing cannot add information to images. For example, if an imaging modality
cannot resolve objects less that a particular size, image processing cannot cause such
objects to be seen.
In some cases, the information of interest is visible in the unprocessed images and
is merely made more conspicuous. In other cases, information of interest may not be
visible at all in the unprocessed image data. X-ray CT is an example of the latter case;
observation of the raw projection data fails to reveal many of the structures visible
in the processed cross-sectional images. The following examples of image processing
are described only superficially and will be discussed in detail in later chapters.
The addition or subtraction of digital images is used in several imaging modalities. Both of these operations require the images being added or subtracted to be of
the same format (same number of pixels along each axis of the images) and produce
a resultant image (sum image or difference image) in that format. In image subtraction, each pixel in one image is subtracted from the corresponding pixel in a second
image to yield the corresponding pixel value in the difference image. Image addition
is similar, but with pixel-by-pixel addition instead of subtraction. Image subtraction
is used in digital subtraction angiography to remove the effects of stationary anatomic
structures not of interest from the images of contrast-enhanced blood vessels and in
nuclear gated cardiac blood pool imaging to yield difference images depicting ventricular stroke-volume and ventricular wall dyskinesis.
Spatial filtering is used in many types of medical imaging. Medical images commonly
have a grainy appearance, called quantum mottle, caused by the statistical nature of the
acquisition process. The visibility of quantum mottle can be reduced by a spatial filtering operation called smoothing. In most spatial smoothing algorithms, each pixel value
in the smoothed image is obtained by a weighted averaging of the corresponding pixel
in the unprocessed image with its neighbors. Although smoothing reduces quantum
mottle, it also blurs the image. Images must not be smoothed to the extent that clinically significant detail is lost. Spatial filtering can also enhance the edges of structures in
an image. Edge-enhancement increases the statistical noise in the image. Convolution
filtering, the method used for most spatial filtering, is discussed in detail in Chapter 4.
A computer can calculate physiologic performance indices from image data. For
example, the estimation of the left ventricular ejection fraction from nuclear gated
cardiac blood pool image sequences is described in Chapter 18. In some cases, these
data may be displayed graphically, such as the time-activity curves of a bolus of a
radiopharmaceutical passing through the kidneys.
In x-ray and nuclear CT, sets of projection images are acquired from different
angles about the long axes of patients. From these sets of projection images, computers calculate cross-sectional images depicting tissue linear attenuation coefficients
(x-ray CT) or radionuclide concentration (SPECT and PET). The methods by which
volumetric data sets are reconstructed from projection images are discussed in detail
in Chapters 10 and 19.
160
In image co-registration, image data from one medical imaging modality is superimposed on a spatially congruent image from another modality. Image co-registration
is especially useful when one of the modalities, such as SPECT or PET, provides
physiologic information lacking anatomic detail and the other modality, often MRI or
x-ray CT, provides inferior physiologic information, but superb depiction of anatomy.
When the images are obtained using two separate devices, co-registration poses formidable difficulties in matching the position, orientation, and magnification of one
image to the other. To greatly simplify co-registration, PET and SPECT devices that
incorporate x-ray CT systems have been developed and are discussed in Chapter 19.
Computer-Aided Detection
Computer-aided detection, also known as computer-aided diagnosis, uses a computer program to detect features likely to be of clinical significance in images. Its
purpose is not to replace the interpreting physician, but to assist by calling attention
to structures that might have been overlooked. For example, software is available to
automatically locate structures suggestive of masses, clusters of microcalcifications,
and architectural distortions in mammographic images (Chapter 8). Computer-aided
detection can improve the sensitivity of the interpretation, but also may reduce the
specificity. (Sensitivity and specificity are defined in Chapter 4.)
Display of Volumetric Image Data

High spatial resolution, thin slice tomographic imaging modalities (e.g., CT and MRI)
provide image sets that can exceed a thousand images covering large volumes of
patients. SPECT and PET, which have lower spatial resolution and therefore typically
produce smaller image sets, are commonly co-registered on CT image sets. The
display and interpretation of these massive image sets presents formidable challenges
to display systems, interpreting physicians, and other clinicians.
Stack versus Tile Mode Display

Tomographic images were once recorded on film and displayed side by side on
motorized light boxes. In early PACS systems, radiologists often replicated this sideby-side viewing by attempting to simultaneously display all images in a study. This
side-by-side image display is referred-to as tile-mode display and early display
workstations were commonly equipped with four or more video monitors for this
purpose. Tile mode display became increasing impractical with thin slice image
acquisitions and correspondingly larger numbers of transaxial images. Tile mode
display has been largely replaced by stack mode display, in which only a single image
is displayed at a time and, as the viewing clinician manipulates a pointing device or
key on the keyboard, the computer successively replaces that image with the next
adjacent image. Stack mode permits the viewer to follow blood vessels and organ
boundaries while scrolling through the images.
Multiplanar Reformatting
Multiplanar reconstruction (MPR) is the reformatting of a volumetric dataset (e.g.,
a stack of axial images) into tomographic images by selecting pixel values from the
dataset that correspond to the desired tomographic image planes. MPR is most
commonly used to produce images that are parallel to one or more of the three
orthogonal planes: axial, coronal, and sagittal. In many cases (especially CT), the
data are acquired in the axial plane. A coronal image can be formed from such a
161
dataset by selecting a specific row of pixel values from each axial image. Similarly,
a sagittal image can be created by selecting a specific column of pixel values from
each axial image. Oblique reconstructions (arbitrary planes that are not orthogonal
to the axial plane), slab (thicker plane) presentations, and curved MPR images
are also commonly produced. The latter are obtained along user-determined curved
surfaces through the volumetric dataset that follow anatomic structures such as the
spine, aorta, or a coronary artery. In MPR, particularly in creating oblique and curved
images, some pixels of the MPR images may not exactly correspond to the locations
of values in the volumetric dataset. In this case, the pixel values in the MPR images
are created by interpolation of the values in the nearest pixels in the volumetric dataset. Figure 5-28 shows three MPR images.
Maximum Intensity Projection Images

Maximum intensity projection (MIP) is a method of forming projection images by
casting rays through a volume dataset and selecting the maximal pixel value along
each ray. The rays may be parallel or they may diverge from a point in space; in either
case, each ray intersects the center of one pixel in the projection image. This method
is extremely useful for depicting high signals in a volume (e.g., contrast CT or MRI
studies) for vascular angiography or demonstrating calcified plaque in vasculature,
for instance. A variation is thin or thick slab MIP, using MPR projections and a selectable number of image slices (a slab), displaying the maximum values obtained along
rays projected through the slab. Figure 5-28 shows two thin-slab MIP images.
Volume Rendering
For some applications, there are advantages to providing a 2D image that contains
depth cues to permit it to be viewed as a 3D object. Two approaches are shaded surface
displays, also called surface rendering (not significantly used) and volume rendering.
In both approaches, the first step is to segment the volume set into different structures
(e.g., bones and soft tissue). Segmentation may be done entirely by the computer or
with guidance by a person. Segmentation is simplest when there are large differences
in the pixel values among the tissues or other objects to be segmented.
Shaded surface display (SSD) provides a simulated 3D view of surfaces of an object or
objects in the volume dataset. First, the surfaces to be displayed must be identified. For
instance, for bony structures on CT images, this can be by simple thresholding to exclude
soft tissues. Alternatively, they may be identified by a gradient edge-detection technique.
Surfaces are identified by marking individual voxels as belonging or not belonging to
the surface, creating a binary dataset. Then the computer calculates the observed light
intensity in the SSD image by calculating the reflections from simulated direct and diffuse
light sources. The surface may be displayed in shades of gray or in color, such as fleshtone for fly-through colonoscopy image sequences. Shaded surface display algorithms
are computationally efficient; however, a major disadvantage of SSDs is that only a very
small fraction of the information in the original image dataset is displayed. Furthermore,
errors may occur in the identification of the surface to be displayed.
Volume rendering, in contrast to SSD and MIP, uses a more complete set of
the voxels within the stacked axial images. In volume rendering, each voxel in the
image volume of a stacked tomographic image set is assigned an opacity ranging
from 0% to 100% and a color. A voxel assigned an opacity of 0% will be invisible in the final image and assigning a voxel an opacity of 100% will prevent the
viewing of voxels behind it. A voxel with an opacity between 0% and 100% will be
visible and will permit the viewing of voxels behind it. The voxels in the volume
image set are segmented into those corresponding to various organs, tissues, and
162
FIGURE 5-28 A volumetric view from a 3D tomographic dataset is depicted. Top. multiplanar reconstruction from axial images include coronal and sagittal views with 1.25 mm slice thickness; Bottom. volume
rendered dataset from the full volume with bone removal, and corresponding thin-slab maximum intensity
projection (MIP) images for the anterior and posterior projections.
objects (e.g., bone, soft tissue, contrast enhanced blood vessels, air, fat) based
upon specific ranges of pixel values (e.g., CT numbers or MRI digital values), perhaps also using assumptions about anatomy and/or guidance by a person. Then,
opacity values and colors are assigned to voxels containing specific tissues. The
assignments of these artificial characteristics to the voxels containing specific tissues are selected from predefined templates for specific imaging protocols and
organ systems (e.g., CT angiogram, pulmonary embolism, or fracture). Next, as
in MIP image formation, a set of rays, parallel or diverging from a point, are cast
through the volume dataset, with each ray passing through the center of a pixel in
the volume rendered image. Each pixel value in the volume rendered image is calculated from the opacities and colors of the voxels along the ray corresponding to
that pixel. Simulated lighting effects may be added. Volume rendering provides a
robust view of the anatomy and depth relationships. On the other hand, it displays
only a small fraction of the information in the initial image dataset, although much
more that SSD or MIP. Furthermore, segmentation errors can occur, causing errors
in the displayed anatomy. Additionally, the color assignments, while appearing
realistic, are arbitrary and can be misleading because of erroneous tissue classification. Nevertheless, volume rendering is useful for surgical planning and provides
images that are easy for referring physicians to understand. Figure 5-28 shows
thin-slab MIP and volume rendered images.
163
5.9 Security, Including Availablility

An important issue regarding medical imaging, PACS, and teleradiology is information security. The main goals of information security are (1) privacy, (2) integrity, (3)
authentication, (4) nonrepudiation, and (5) availability. Privacy, also called confidentiality, refers to denying persons, other than intended recipients, access to confidential information. Integrity means that information has not been altered, either
accidentally or deliberately. Authentication permits the recipient of information to
verify the identity of the sender and the sender to verify the identity of the intended
recipient. Nonrepudiation prevents a person from later denying an action, such as
approval of a payment or approval of a report. Availability means that information
and services are available when needed.
Privacy is commonly achieved either by storing information in a secure location,
such as locked room, or by encrypting the information into a code that only intended
recipients can convert to its original form. Unintentional alteration of information
can be detected by use of an error detecting code. A sending computer calculates the
error detecting code from the information to be transmitted over a network and sends
the code along with the transmitted information. The receiving computer calculates
the error detecting code from the information it receives and compares the code it
calculated to the code sent with information. If the two codes match, there is a high
degree of assurance that the information was transmitted without error. Protection
against intentional modification can be achieved by encrypting both the transmitted
information and the error detecting code. Authentication is usually performed by
the use of passwords, which are encrypted when transmitted across a network, or by
using biometric methods, such as a scan of a persons fingerprint or retina. Measures
to ensure availability are described later in this section.
The federal Health Insurance Portability and Accountability Act of 1996 (HIPAA)
and associated federal regulations (45 CFR 164) impose security requirements for
electronic protected healthcare information. Goals of the HIPAA regulations are to
1. Ensure the confidentiality, integrity, and availability of all electronic protected
healthcare information
2. Identify and protect against reasonably anticipated threats or hazards to the
security or integrity of the information
3. Protect against reasonably anticipated impermissible uses or disclosures
4. Ensure compliance by the workforce
The security measures described below largely conform to HIPAA security regulations, but do not include all such requirements. Furthermore, the regulations are
subject to change. Hence, a person responsible for a security program should refer
to the current regulations.
A major danger in using a computer is that the operating system, applications
programs, and important information are often stored on a single disk; an accident
could cause all of them to be lost. The primary threats to information and software
on digital storage devices or media are mechanical or electrical malfunction, such
as a disk head crash; human error, such as the accidental deletion of a file or the
accidental formatting of a disk (potentially causing the loss of all information on
the disk); and malicious damage. To reduce the risk of information loss, important
files should be copied (backed up) onto magnetic disks, optical disks, or magnetic tape at regularly scheduled times, with the backup copies stored in a distant
secure location.
164
Programs written with malicious intent are a threat to computers. The most prevalent of these are computer viruses. A virus is a string of instructions hidden in a program. If a program containing a virus is loaded into a computer and executed, the virus
copies itself into other programs stored on mass storage devices. If a copy of an infected
program is sent to another computer and executed, that computer becomes infected.
Although a virus need not do harm, many deliberately cause damage, such as the deletion of all files on the disk on Friday the 13th. Viruses that are not intended to cause
damage may interfere with the operation of the computer or cause damage because they
are poorly written. A computer cannot be infected with a virus by the importation of
data alone or by the user reading an e-mail message. However, a computer can become
infected if an infected file is attached to an e-mail message and if that file is executed.
Malicious programs also include Trojan horses, programs that are presented as
being of interest so people will load them onto computers, but have hidden purposes;
worms, programs that automatically spread over computer networks; time bombs,
programs or program segments that do something harmful, such as change or delete
information, on a future date; key loggers, programs that record everything typed by
a user, which can later be viewed for information of use, such as login names, passwords, and financial and personal information; and password grabbers, programs
that ask people to log in and store the passwords and other login information for
unauthorized use. A virus, worm, or Trojan horse may incorporate a time bomb or
key logger. In 1988, a worm largely brought the Internet to a standstill. The primary
way to reduce the chance of a virus infection or other problem from malicious software is to establish a policy forbidding the loading of storage media and software from
untrustworthy sources. Commercial virus-protection software, which searches files
on storage devices and files received over a network for known viruses and removes
them, should be used. The final line of defense, however, is the saving of backup copies. Once a computer is infected, it may be necessary to reformat all disks and reload
all software and information from the backup copies. A related threat is a denial of
service attack, the bombardment of a computer or network with so many messages
that it cannot function. Denial of service attacks are commonly launched from multiple computers that are being controlled by malicious software.
Sophisticated computer operating systems provide security features including
password protection and the ability to grant individual users different levels of access
to stored files. Measures should be taken to deny unauthorized users access to your
system. Passwords should be used to deny access, directly and over a network or
a modem. Each user should be granted only the privileges required to accomplish
needed tasks. For example, technologists who acquire and process studies and
interpreting physicians should not be granted the ability to delete or modify system
software files or patient studies, whereas the system manager must be granted full
privileges to all files on the system.
Security measures include administrative, physical, and technical safeguards.
The following is a list of elements in a PACS and teleradiology security plan:
1. Perform a risk analysis.
2. Establish written policies and procedures for information security.
3. Train staff in the policies and procedures.
4. BackupsMaintain copies of important programs and information in case the
data on a single device are lost. Backup copies should be stored in a secure location remote from the primary storage.
5. Install commercial anti-virus software on all computers to identify and remove
common viruses and periodically update the software to recognize the signatures
of recent viruses.
165
6. Forbid the loading of removable media (e.g., floppy and optical disks from the
homes of staff) and programs from nontrustworthy sources and forbid activating
attachments to unexpected e-mail messages.
7. Authenticate users, directly and over a network or a modem, by passwords. Use
secure passwords.
8. Terminate promptly the access privileges of former employees.
9. Log off workstations, particularly those in nonsecure areas, when not in use.
10. Grant each user only sufficient privileges required to accomplish needed tasks
and only access to information to which the user requires access.
11. Secure transferEncrypt information transferred over nonsecure networks.
12. Secure storagePhysically secure media (e.g., store it in a room to which access
is controlled) or encrypt the information on it.
13. Erase information on or destroy removable media and storage devices before
disposal or transfer for reuse. On most operating systems, deleting a file merely
removes the listing of the file from the device directory. The information remains
stored on the device or media.
14. Install patches to the operating system to fix security vulnerabilities.
15. Install firewalls (described below) at nodes where your LAN is connected to
other networks.
16. Audit trailsEach access to protected healthcare information must be
recorded.
17. Establish an emergency operations and disaster recovery plan.
The goal of availability is to ensure that acquired studies can be interpreted and
stored images can be viewed at all times. Availability is achieved by the use of reliable components and media, fault-tolerant and risk-informed design and installation,
provisions for prompt repair, and an emergency operation and disaster recovery plan.
Adesign that continues to function despite equipment failure is said to be fault tolerant. The design of a PACS must take into account the fact that equipment will fail. Fault
tolerance usually implies redundancy of critical components. Not all components are
critical. For example, in a PACS with a central archive connected by a network to multiple workstations, the failure of a single workstation would have little adverse effect,
but failure of the archive or network could prevent the interpretation of studies.
An essential element of fault tolerance is for the PACS to create and maintain copies of all information on separate and remote storage devices for backup and disaster
recovery. Offsite storage for disaster recovery, perhaps leased from a commercial vendor, is an option. The value of a study declines with time after its acquisition and so,
therefore, does the degree of protection required against its loss. A very high degree
of protection is necessary until it has been interpreted.
The design of the PACS and the policies and procedures for its use should take
into account the probability and possible consequences of risks, such as human error;
deliberate attempts at sabotage, including computer viruses, worms, and denial of
service attacks; fire; water leaks; and natural disasters such as floods, storms, and
earthquakes. Examples are installing a redundant PACS archive far from the primary
archive, preferably in another building, and, in an area vulnerable to flooding, not
installing a PACS archive in a basement or ground floor of a building.
Provisions for repair of a PACS are important. A failure of a critical component is
less serious if it is quickly repaired. Arrangements for emergency service should be
made in advance of equipment failure.
Computer networks pose significant security challenges. Unauthorized persons
can access a computer over the network and, on some LANs, any computer on the
166
network can be programmed to read the traffic. If a network is connected to the

Internet, it is vulnerable to attack by every hacker on the planet.
Not only must the confidentiality and integrity of information be protected during transmission across a network, but the computers on the network must be protected from unauthorized access via the network. A simple way to protect a network
is to not connect it to other networks. However, this may greatly limit its usefulness.
For example, at a medical center, it is useful to connect the network supporting a
PACS to the main hospital network. The main hospital network typically provides a
link to the Internet as well.
A firewall can enhance the security of a network or network segment. A firewall is
a program, router, computer, or even a small network that is used to connect two networks to provide security. There are many services that can be provided by a firewall.
One of the simplest is packet filtering, in which the firewall examines packets reaching it. It reads their source and destination addresses and the applications for which
they are intended and, based upon rules established by the network administrator,
forwards or discards them. For example, packet filtering can limit which computers
a person outside the firewall can access and can forbid certain kinds of access, such
as FTP and Telnet. Different rules may be applied to incoming and outgoing packets.
Thus, the privileges of users outside the firewall can be restricted without limiting
the ability of users on the protected network to access computers past the firewall.
Firewalls also can maintain records of the traffic across the firewall, to help detect
and diagnose attacks on the network. A firewall, by itself, does not provide complete
protection and thus should merely be part of a comprehensive computer security
program.
SUGGESTED READINGS
American Association of Physicists in Medicine Task Group 18, AAPM On-Line Report No. 03, Assessment
of display performance for medical imaging systems. April 2005.
American College of Radiology, ACR Technical Standard for Electronic Practice of Medical Imaging. ACR
Practice Guidelines and Technical Standards, 2007.
American College of Radiology. ACRSIIM Practice Guideline for Electronic Medical Information Privacy
and Security. ACR Practice Guidelines and Technical Standards, 2009.
Badano A. Principles of cathode ray tube and liquid crystal display devices. In: Advances in digital
radiography: categorical course in diagnostic radiology physics. Oak Brook, IL: Radiological Society of
North America, 2003:91102.
Barten PGJ. Contrast sensitivity of the human eye and its effects on image quality. Bellingham, WA: SPIE
Optical Engineering Press, 1999.
Branstetter BF IV. Basics of imaging informatics, part 1. Radiology 2007;243:656667.
Clunie DA. DICOM implementations for digital radiography. In: Advances in digital radiography: categorical
course in diagnostic radiology physics. Radiological Society of North America, 2003:163172.
Code of Federal Regulations, 45 CFR Part 164, HIPAA Privacy and Security Regulations.
Cohen MD, Rumreich LL, Garriot KM, et al. Planning for PACS: a comprehensive guide to nontechnical
considerations. J Am Coll Radiol 2005;2:327337.
Cusma JT, Wondrow MA, Holmes DR. Image storage considerations in the cardiac catheterization
laboratory. In: Categorical course in diagnostic radiology physics: cardiac catheterization imaging. Oak
Brook, IL: Radiological Society of North America, 1998.
Digital Imaging and Communications in Medicine (DICOM) Part 1: Introduction and Overview, PS
3.1-2011, National Electrical Manufacturers Association, 2011.
Digital Imaging and Communications in Medicine (DICOM) Part 14: Grayscale Standard Display Function,
PS 3.14-2011, National Electrical Manufacturers Association, 2011.
Flynn MG. Visual requirements for high-fidelity display. In: Advances in digital radiography: categorical
course in diagnostic radiology physics. Radiological Society of North America, 2003:103107.
Myers RL. Display interfaces: Fundamentals and Standards. John Wiley & Sons, Ltd, Chichester, UK,
2003.
167
Seibert JA, Filipow LJ, Andriole KP, eds. Practical digital imaging and PACS. Madison, WI: Medical Physics
Publishing, 1999.
Tanenbaum AS, Wetherall DJ. Computer networks. 5th ed. Upper Saddle River, NJ: Prentice Hall PTR,
2010.
Wallis JW, Miller TR, Three-dimensional display in nuclear medicine and radiology. J Nucl Med 1991;32:
534546.
SELECTED REFERENCES
Goo JM, Choi JY, et al. Effect of Monitor Luminance and Ambient Light on Observer Performance in
Soft-Copy Reading of Digital Chest Radiographs. Radiology, September 2004, Volume 232 Number
3: 762-766
Krupinski EA. Technology and Perception in the 21st-Century Reading Room. J Am Coll Radiol 2006;3:
433-440.
Section
II
Diagnostic Radiology
Chapter
x-ray Production, X-ray Tubes,

and x-ray Generators
x-rays are produced when highly energetic electrons interact with matter, converting
some of their kinetic energy into electromagnetic radiation. A device that produces
x-rays in the diagnostic energy range typically contains an electron source, an evacuated
path for electron acceleration, a target electrode, and an external power source to provide
a high voltage (potential difference) to accelerate the electrons. Specifically, the x-ray tube
insert contains the electron source and target within an evacuated glass or metal envelope; the tube housing provides protective radiation shielding and cools the x-ray tube
insert;the x-ray generator supplies the voltage to accelerate the electrons; x-ray beam
filters at thetube port shape the x-ray energy spectrum; and collimators define the size
and shape of the x-ray field incident on the patient. The generator also permits control
of the x-ray beam characteristics through the selection of voltage, current, and exposure
time. These components work in concert to create a beam of x-ray photons of welldefined intensity, penetrability, and spatial distribution. In this chapter, the x-ray creation
process, characteristics of the x-ray beam, and equipment components are discussed.
6.1 Production of x-rays
Bremsstrahlung Spectrum
x-rays are created from the conversion of kinetic energy of electrons into electromagnetic
radiation when they are decelerated by interaction with a target material. A simplified
diagram of an x-ray tube (Fig. 6-1) illustrates these components. For diagnostic radiology, a large electric potential difference (the SI unit of potential difference is the volt, V)
of 20,000 to 150,000 V (20 to 150 kV) is applied between two electrodes (the cathode
and the anode) in the vacuum. The cathode is the source of electrons, and the anode, with
a positive potential with respect to the cathode, is the target of electrons. As electrons
from the cathode travel to the anode, they are accelerated by the voltage between the
electrodes and attain kinetic energies equal to the product of the electrical charge and
potential difference (see Appendix A). Acommon unit of energy is the electron volt (eV),
equal to the energy attained by an electron accelerated across a potential difference of 1
V. Thus, the kinetic energy of an electron accelerated by a potential difference of 50kV is
50 keV. One eV is a very small amount of energy, as there are 6.24 1018 eV/j.
On impact with the target, the kinetic energy of the electrons is converted to
other forms of energy. The vast majority of interactions are collisional, whereby energy
exchanges with electrons in the target give rise to heat. A small fraction of the accelerated electrons comes within the proximity of an atomic nucleus and is influenced by its
positive electric field. As discussed in Chapter 3, electrical (Coulombic) forces attract
and decelerate an electron and change its direction, causing a loss of kinetic energy,
which is emitted as an x-ray photon of equal energy (i.e., bremsstrahlung r adiation).
171
172
Section II Diagnostic Radiology

Electrons
Tungsten Target
Evacuated Envelope
Copper Anode
Heated Tungsten Filament

Cathode
voltage
voltage
+
High Voltage
Source
X-rays
FIGURE 6-1 Minimum requirements for x-ray production include a source and target of electrons, an evacuated envelope, and connection of the electrodes to a high-voltage source.
The amount of energy lost by the electron and thus the energy of the resulting x-ray are determined by the distance between the incident electron and
the target nucleus, since the Coulombic force is proportional to the inverse
of the square of the distance. At relatively large distances from the nucleus,
the Coulombic attraction is weak; these encounters produce low x-ray energies
(Fig. 6-2, electron no. 3). At closer interaction distances, the force acting on
Target atom
Nucleus
Incident electrons
1
+
+++
2
3
Close interaction:
Moderate energy
3
1
Distant interaction:
Low energy
Impact with nucleus:

Maximum energy
FIGURE 6-2 Bremsstrahlung radiation arises from energetic electron interactions with an atomic nucleus of
the target material. In a close approach, the positive nucleus attracts the negative electron, causing deceleration and redirection, resulting in a loss of kinetic energy that is converted to an x-ray. The x-ray energy depends
on the interaction distance between the electron and the nucleus; it decreases as the distance increases.
Chapter 6 x-ray Production, X-ray Tubes, and x-ray Generators
173
the electron increases, causing a greater deceleration; these encounters produce

higher x-ray energies (see Fig. 6-2, electron no. 2). A nearly direct impact of
an electron with the target nucleus results in loss of nearly all of the electrons
kinetic energy (see Fig. 6-2, electron no. 1). In this rare situation, the highest
x-ray energies areproduced.
The probability of electron interactions that result in production of x-ray
energy E is dependent on the radial interaction distance, r, from the nucleus,
which defines a circumference, 2pr. With increasing distance from the nucleus,
the circumference increases, and therefore the probability of interaction increases,
but the x-ray energy decreases. Conversely, as the interaction distance, r, decreases,
the x-ray energy increases because of greater electron deceleration, but the probability of interaction decreases. For the closest electron-atomic nuclei interactions,
the highest x-ray energies are produced. However, the probability of such an interaction is very small, and the number of x-rays produced is correspondingly small.
The number of x-rays produced decreases linearly with energy up to the maximal
x-ray energy, which is equal to the energy of the incident electrons. Abremsstrahlung spectrum is the probability distribution of x-ray photons as a function of photon energy (keV). The unfiltered bremsstrahlung spectrum (Fig. 6-3A) shows an
inverse linear relationship between the number and the energy of the x-rays produced, with the highest x-ray energy determined by the peak voltage (kV) applied
across the x-ray tube. A typical filtered bremsstrahlung spectrum (Fig. 6-3B) has
no x-rays below about 10 keV; the numbers increase to a maximum at about one
third to one half the maximal x-ray energy and then decrease to zero as the x-ray
energy increases to the maximal x-ray energy. Filtration in this context refers to the
removal of x-rays by attenuation in materials that are inherent in the x-ray tube
(e.g., the glass window of the tube insert), as well as by materials that are purposefully placed in the beam, such as thin aluminum and copper sheets, to remove
lower energy x-rays and adjust the spectrum for optimal low-dose imaging (see
Section 6.7).
Major factors that affect x-ray production efficiency include the atomic number of the target material and the kinetic energy of the incident electrons. The
approximate ratio of radiative energy loss caused by bremsstrahlung production to
90 kVp operational voltage
Relative output
(a) Unfiltered bremsstrahlung

spectrum
(b) Filtered bremsstrahlung
spectrum
90 keV maximal
photon energy
(c)
10
20
30
40
50
60
Energy (keV)
70
80
90
FIGURE 6-3 The bremsstrahlung

energy distribution for a 90-kV acceleration potential difference. The
unfiltered bremsstrahlung spectrum
(a) shows a greater probability of
low-energy x-ray photon production that is inversely linear with
energy up to the maximum energy
of 90 keV. The filtered spectrum (b)
shows the preferential attenuation
of the lowest-energy x-ray photons.
The vertical arrow (c) indicates the
average energy of the spectrum,
which is typically 1/3 to 1/2 the
maximal energy.
174
c ollisional (excitation and ionization) energy loss within the diagnostic x-ray energy
range (potential difference of 20 to 150 kV) is expressed as follows:
Radiative energy loss
E KZ
Collisional energy loss 820,000
[6-1]
where EK is the kinetic energy of the incident electrons in keV, and Z is the atomic
number of the target electrode material. The most common target material is tungsten (W, Z = 74); in mammography, molybdenum (Mo, Z = 42) and rhodium (Rh,
Z = 45) are also used. For 100-keV electrons impinging on tungsten, the approximate ratio of radiative to collisional losses is (100 74)/820,000 0.009 0.9%;
therefore, more than 99% of the incident electron energy on the target electrode
is converted to heat and nonuseful low-energy electromagnetic radiation. At much
higher electron energies produced by radiation therapy systems (millions of electron volts), the efficiency of x-ray production is dramatically increased. However,
Equation 6-1is not applicable beyond diagnostic imaging x-ray energies.
Characteristic x-ray Spectrum

In addition to the continuous bremsstrahlung x-ray spectrum, discrete x-ray energy
peaks called characteristic radiation can be present, depending on the elemental
composition of the target electrode and the applied x-ray tube voltage. Electrons in
an atom are distributed in shells, each of which has an electron binding energy. The
innermost shell is designated the K shell and has the highest electron binding energy,
followed by the L, M, and N shells, with progressively less binding energy. Table 6-1
lists the common anode target materials and the corresponding binding energies of
their K, L, and M electron shells. The electron binding energies are characteristic
of the elements. When the energy of an incident electron, determined by the voltage
applied to the x-ray tube, exceeds the binding energy of an electron shell in a target
atom, a collisional interaction can eject an electron from its shell, creating a vacancy.
As discussed in Chapter 2, an outer shell electron with less binding energy immediately transitions to fill the vacancy, and a characteristic x-ray is emitted with an energy
equal to the difference in the electron binding energies of the two shells (Fig. 6-4).
For tungsten, an L-shell (binding energy = 10.2 keV) electron transition to fill a
K-shell (binding energy = 69.5 keV) vacancy produces a characteristic x-ray with a
discrete energy of
EbK EbL 69.5 keV 10.2 keV 59.3 keV
Electron transitions occur from adjacent and nonadjacent electron shells in the
atom, giving rise to several discrete characteristic energy peaks superimposed on the
TABLE 6-1ELECTRON BINDING ENERGIES (keV)
OF COMMON x-ray TUBE TARGET

MATERIALS
ELECTRON SHELL
TUNGSTEN
MOLYBDENUM
69.5
20.0
RHODIUM
23.2
12.1/11.5/10.2
2.8/2.6/2.5
3.4/3.1/3.0
2.81.9
0.50.4
0.60.2

Ejected K-shell electron
Target atom
Nucleus
+
+++
Rebounding
incident electron
175
FIGURE 6-4 Generation of a characteristic x-ray in a target atom occurs

in the following sequence: (1) The incident electron interacts with the K-shell
electron via a repulsive electrical force.
(2) The K-shell electron is removed
(only if the energy of the incident electron is greater than the K-shell binding energy), leaving a vacancy in the
K-shell. (3) An electron from the adjacent L-shell (or possibly a different shell)
fills the vacancy. (4) A Ka characteristic
x-ray photon is emitted with energy
equal to the difference between the
binding energies of the two shells. In
this case, a 59.3-keV photon is emitted.
Characteristic X-ray
bremsstrahlung spectrum. Characteristic x-rays are designated by the shell in which

the electron vacancy is filled, and a subscript of a or b indicates whether the electron
transition is from an adjacent shell (a) or nonadjacent shell (b). For example, Ka
refers to an electron transition from the L to the K shell, and Kb refers to an electron
transition from the M, N, or O shell to the K shell. A Kb x-ray is more energetic than
a Ka x-ray. Characteristic x-rays other than those generated by K-shell transitions are
too low in energy for any useful contributions to the image formation process and are
undesirable for diagnostic imaging. Table 6-2 lists electron shell binding energies and
corresponding K-shell characteristic x-ray energies of W, Mo, and Rh anode targets.
Characteristic K x-rays are produced only when the electrons impinging on the
target exceed the binding energy of a K-shell electron. x-Ray tube voltages must therefore be greater than 69.5 kV for W targets, 20 kV for Mo targets, and 23 kV for Rh
targets to produce K characteristic x-rays. In terms of intensity, as the x-ray tube voltage
increases, so does the ratio of characteristic to bremsstrahlung x-rays. For example, at
80 kV, approximately 5% of the total x-ray output intensity for a tungsten target is composed of characteristic radiation, which increases to about 10% at 100kV. Figure6-5
illustrates a bremsstrahlung plus characteristic radiation spectrum. In mammography,
characteristic x-rays from Mo and Rh target x-ray tubes are particularly useful in optimizing image contrast and radiation dose (See Chapter 8 for further information).
TABLE 6-2K-SHELL CHARACTERISTIC X-RAY
NERGIES (keV) OF COMMON X-RAY

E
TUBETARGET MATERIALS
SHELL TRANSITION
TUNGSTEN
MOLYBDENUM
RHODIUM
Ka1
59.32
17.48
20.22
Ka2
57.98
17.37
20.07
Kb1
67.24
19.61
22.72
Note: Only prominent transitions are listed. The subscripts 1 and 2 represent
energy levels that exist within each shell.
FIGURE 6-5 The filtered spectrum

of bremsstrahlung and characteristic
radiation from a tungsten target with
a potential difference of 90 kV illustrates specific characteristic radiation
energies from Ka and Kb transitions.
Filtration (the preferential removal of
low-energy photons as they traverse
matter) is discussed in Section 6.5.
Characteristic
Radiation Spikes
(Tungsten)
K1
Relative output
176
K2
10
20
30
K1
40
50
60
Energy (keV)
70
80
90
6.2 x-ray Tubes

The x-ray tube provides an environment for the production of bremsstrahlung and
characteristic x-rays. Major tube components are the cathode, anode, rotor/stator, glass
or metal envelope, tube port, cable sockets, and tube housing, illustrated in Figure 6-6.
An actual x-ray tube showing the x-ray tube insert and part of the housing is shown
in Figure 6-7. The x-ray generator (Section 6.3) supplies the power and permits
selection of tube voltage, tube current, and exposure time. Depending upon the type
of imaging examination and the characteristics of the anatomy being imaged, the
x-ray tube voltage is set to values from 40 to 150 kV for diagnostic imaging, and 25
to 40 kV for mammography. The x-ray tube current, measured in milliamperes (mA),
is proportional to the number of electrons per second flowing from the cathode to
the anode, where 1 mA = 6.24 1015 electrons/s. For continuous fluoroscopy, the
tube current is relatively low, from 1 to 5 mA, and for projection radiography, the
tube current is set from 50 to 1,200 mA in conjunction with short exposure times
(typically less than 100 ms). (In pulsed fluoroscopy, the tube current is commonly
delivered in short pulses instead of being continuous; the average tube current is
typically in the range of 10 to 50 mA, while the overall number of electrons delivered through the tube is about the same per image.) The kV, mA, and exposure time
are the three major selectable parameters on the x-ray generator control panel that
FIGURE 6-6 A diagram of the
major components of a modern
x-ray tube and housing assembly is
shown.
Cable sockets
X-ray tube insert
Expansion
bellows
Anode rotor
Transformer oil
Lead shielding
Stator windings
Anode
Cathode
Output port
177
X-ray tube
insert
Housing corver
Stator
windings
Lead
shielding
Cathode
Tube port
Anode
FIGURE 6-7 Picture of an x-ray tube insert and partially cut-away housing, shows the various components
of the x-ray tube. For this housing, the lead shielding thickness is 2 mm.
determine the x-ray beam characteristics. Often, the product of the tube current and
exposure time are considered as one entity, the mAs (milliampere-second; technically, mAs is a product of two units but, in common usage, it serves as a quantity).
These parameters are discussed further in the following sections.
Cathode
The cathode is the negative electrode in the x-ray tube, comprised of a filament or
filaments and a focusing cup (Fig. 6-8). A filament is made of tungsten wire wound in
a helix, and is electrically connected to the filament circuit, which provides a voltage
of approximately 10 V and variable current up to 7,000 mA (7 A). Most x-ray tubes
for diagnostic imaging have two filaments of different lengths, each positioned in a
slot machined into the focusing cup, with one end directly connected to the focusing
cup, and the other end electrically insulated from the cup by a ceramic insert. Only
one filament is energized for an imaging examination. On many x-ray systems, the
small or the large filament can be manually selected, or automatically selected by the
x-ray generator depending on the technique factors (kV and mAs).
When energized, the filament circuit heats the filament through electrical resistance, and the process of thermionic emission releases electrons from the filament
surface at a rate determined by the filament current and corresponding filament
temperature. Heat generated by resistance to electron flow in the filament raises the
temperature to a point where electrons can leave the surface. However, electrons
flow from the cathode to the anode only when the tube voltage is applied between these
electrodes. The numbers of electrons that are available are adjusted by the filament
current and filament temperature, as shown in Figure 6-9, where small changes in
the filament current can produce relatively large changes in tube current. Output
178
FIGURE 6-8 The x-ray tube

cathode structure consists of the filaments and the focusing (or cathode)
cup. Current from the filament circuit
heats a filament, which releases electrons by thermionic emission.
Top view
Focusing cup
Small focal spot

filament
Large focal spot

filament
Focusing cup
Side view
Filament Circuit:
10 V
3 to 7A
of the x-ray tube is emission-limited, meaning that the filament current determines
the x-ray tube current; at any kV, the x-ray flux is proportional to the tube current.
Higher tube voltages also produce slightly higher tube current for the same filament
current. A filament current of 5 A at a tube voltage of 80 kV produces a tube current
of about 800 mA, whereas the same filament current at 120 kV produces a tube current of about 1,100 mA.
In most x-ray tubes, the focusing cup is maintained at the same potential difference as the filament relative to the anode, and at the edge of the slot, an electric field
exists that repels and shapes the cloud of emitted electrons from the filament surface.
As a large voltage is applied between the cathode and anode in the correct polarity,
electrons are accelerated into a tight distribution and travel to the anode, striking a
small area called the focal spot (Fig. 6-10). The focal spot dimensions are determined
by the length of the filament in one direction and the width of electron distribution
in the perpendicular direction.
A biased x-ray tube has a focusing cup totally insulated from the filament wires
so that its voltage is independent of the filament. Because high voltages are applied
80 kV
1.6
Anode (tube) current (A)
FIGURE 6-9 Relationship of tube

current to filament current for various
tube voltages shows a dependence
of approximately kV1.5. For tube voltages 40 kV and lower, a space charge
cloud shields the electric field so that
further increases in filament current
do not increase the tube current. This
is known as space chargelimited
operation. Above 40 kV, the filament current limits the tube current;
this is known as emission-limited
operation.
120 kV
1.4
60 kV
1.2
40 kV
1.0
Space charge
limited
0.8
0.6
0.4
20 kV
0.2
0
5
Filament current (A)
Electric field voltage, V

(relative to focusing cup)
Focal spot
distribution
No electron flow with

large negative
voltage on focusing cup
700
600
700
600
500
400
300
200
100
0
100
Electron
trajectories
500
400
Electron
potential
lines
300
200
100
0
179
Focusing
cup
700
600
500
400
300
200
100
0
1000
4000
Filament
Unbiased
Bias
Voltage
Supply
Biased
Grid bias
+
Voltage
Supply
Grid Pulsed
FIGURE 6-10 The focusing cup shapes the electron distribution when it is at the same voltage as the
filament (left). Isolation of the focusing cup from the filament and application of a negative bias voltage
(~ 100 V) reduces electron distribution further by increasing the repelling electric fields surrounding the filament and modifying the electron trajectories (middle). At the top are typical electron distributions incident on
the target anode (the focal spot) for the unbiased and biased focusing cups. Application of 4,000 V on an
isolated focusing cup completely stops electron flow, even with high voltage applied on the tube; this is known
as a grid biased or grid pulsed tube (right).
to the cathode, electrical insulation of the focusing cup and the bias supply voltage
is necessary, and can add significant expense to the x-ray system. A voltage of about
100 V negative is applied with respect to the filament voltage to further reduce the
spread of electrons and produce a smaller focal spot width (Fig. 6-10 middle). Even
greater negative applied voltage (about 4,000 V) to the focusing cup actually stops
the flow of electrons, providing a means to rapidly switch the x-ray beam on and off
(Fig. 6-10 right); a tube with this capability is referred to as a grid-biased x-ray tube.
Grid-biased x-ray tube switching is used by more expensive fluoroscopy systems
for pulsed fluoroscopy and angiography to rapidly and precisely turn on and turn
off the x-ray beam. This eliminates the build-up delay and turnoff lag of x-ray tubes
switched at the generator, which cause motion artifacts and produce lower average
x-ray energies and unnecessary patient dose.
Ideally, a focal spot would be a point, eliminating geometric blurring. However,
such a focal spot is not possible and, if it were, would permit only a tiny tube current. In practice, a finite focal spot area is used with an area large enough to permit a
sufficiently large tube current and short exposure time. For magnification studies, a
small focal spot is necessary to limit geometric blurring and achieve adequate spatial
resolution (see Figure 6-16 and Chapter 7 on magnification).
Anode
The anode is a metal target electrode that is maintained at a large positive potential
difference relative to the cathode. Electrons striking the anode deposit most of their
energy as heat, with only a small fraction emitted as x-rays. Consequently, the production of x-rays, in quantities necessary for acceptable image quality, generates a
large amount of heat in the anode. To avoid heat damage to the x-ray tube, the rate
of x-ray production (proportional to the tube current) and, at large tube currents,
180
the duration of x-ray production, must be limited. Tungsten (W, Z = 74) is the most
widely used anode material because of its high melting point and high atomic number. A tungsten anode can handle substantial heat deposition without cracking or
pitting of its surface. An alloy of 10% rhenium and 90% tungsten provides added
resistance to surface damage. Tungsten provides greater bremsstrahlung production
than elements with lower atomic numbers (Equation 6-1).
Molybdenum (Mo, Z = 42) and rhodium (Rh, Z = 45) are used as anode materials
in mammographic x-ray tubes. These materials provide useful characteristic x-rays
for breast imaging (see Table 6-2). Mammographic tubes are described further in
Chapter 8.
Anode Configurations: Stationary and Rotating

A simple x-ray tube design has a stationary anode, consisting of a tungsten insert
embedded in a copper block (Fig. 6-11). Copper serves a dual role: it mechanically
supports the insert and efficiently conducts heat from the tungsten target. However,
the small area of the focal spot on the stationary anode limits the tube current and
x-ray output that can be sustained without damage from excessive temperature. Dental x-ray units and some low-output mobile x-ray machines and mobile fluoroscopy
systems use fixed anode x-ray tubes.
Rotating anodes are used for most diagnostic x-ray applications, mainly because
of greater heat loading and higher x-ray intensity output. This design spreads the
heat over a much larger area than does the stationary anode design, permitting much
larger tube currents and exposure durations. The anode is a beveled disk mounted
on a rotor assembly supported by bearings in the x-ray tube insert (Fig. 6-12). The
rotor consists of copper bars arranged around a cylindrical iron core. A donut-shaped
stator device, comprised of electromagnets, surrounds the rotor and is mounted outside of the x-ray tube insert. Alternating current (AC), the periodic reversal of electron movement in a conductor, passes through the stator windings and produces a
rotating magnetic field (see electromagnetic induction, Section 6.3). This induces an
electrical current in the rotors copper bars, which creates an opposing magnetic field
that causes it to spin. Rotation speeds are 3,000 to 3,600 (low speed) or 9,000 to
10,000 (high speed) revolutions per minute (rpm). X-ray systems are designed such
that the x-ray tube will not be energized if the anode is not at full speed; this is the
cause for the short delay (1 to 2 s) when the x-ray tube exposure button is pushed.
Rotor bearings are heat sensitive and are often the cause of x-ray tube failure.
Bearings require special heat insensitive, nonvolatile lubricants because of the vacuum inside the x-ray tube insert and also require thermal insulation from the anode,
achieved by using a molybdenum (a metal with poor heat conductivity) stem attaching the anode to the rotor. Most rotating anodes are cooled by infrared radiation
emission, transferring heat to the x-ray tube insert and to the surrounding oil bath
FIGURE 6-11 The anode of a
fixed anode x-ray tube consists of a
tungsten insert mounted in a copper block. Heat is removed from
the tungsten target by conduction
into the copper block.
Evacuated
glass housing
Cathode
Copper electrode
Tungsten target
Stator
Anode disk
Rotor
Support
Spindle
Molybdenum stem
181
FIGURE 6-12 The anode of a rotating anode x-ray tube is a tungsten

disk mounted on a bearing-supported
rotor assembly (front view, top left;
side view, top right). The rotor consists of a copper and iron laminated
core and forms part of an induction
motor. The other component is the
stator, which exists outside of the
insert, top right. A molybdenum
stem (molybdenum is a poor heat
conductor) connects the rotor to the
anode to reduce heat transfer to the
rotor bearings (bottom).
Bearings
and tube housing. In imaging situations demanding higher heat loads and more rapid
cooling, such as interventional fluoroscopy and computed tomography (CT), sophisticated designs with externally mounted bearings and oil or water heat exchangers
are employed (see special x-ray tube designs in this section).
The focal track area of the rotating anode is approximately equal to the product of
the circumferential track length (2pr) and the track width (Dr), where r is the radial
distance from the axis of the x-ray tube to the center of the track (Fig. 6-13). Thus,
a rotating anode with a 5-cm focal track radius and a 1-mm track width provides a
focal track with an annular area 314 times greater than that of a fixed anode with a
focal spot area of 1 1 mm. The allowable instantaneous heat loading depends on
Side view of anode and cathode
Anode Angle
Front view of anode

Focal track
Rotating Anode
Cathode
Rotation
Length
Focal area
Focal spot true length
Width
Focal spot effective length
Central Axis
Projected focal spot effective size

Looking down central axis
FIGURE 6-13 The anode (target) angle, , is defined as the angle of the target surface in relation to the
central ray. The focal spot length, as projected down the central axis, is foreshortened, according to the line
focus principle (lower right).
182
the anode rotation speed and the focal spot area. Faster rotation distributes the heat
load over a greater portion of the focal track area for short exposure times. A larger
focal spot allows a greater x-ray beam intensity but causes a loss of spatial resolution
that increases with distance of the imaged object from the image receptor. A large
focal spot, which permits high x-ray output and short exposure times, should be
used in situations where motion is expected to be a problem and geometric magnification is small (the object is close to the image receptor).
Anode Angle, Field Coverage, and Focal Spot Size

The anode angle is defined as the angle of the target surface with respect to the
central ray (central axis) in the x-ray field (Fig. 6-13, left diagram). Anode angles in
diagnostic x-ray tubes typically range from 7 to 20 degrees, with 12- to 15-degree
angles being most common. Major factors affected by the anode angle include the
effective focal spot size, tube output intensity, and x-ray field coverage provided at a
given focal spot to detector distance.
The actual focal spot size is the area on the anode that is struck by electrons, and
is primarily determined by the length of the cathode filament and the width of the
focusing cup slot. However, the projected length of the focal spot area at the x-ray
field central ray is much smaller, because of geometric foreshortening of the distribution from the anode surface. Thus, the effective and actual focal spot lengths are
geometrically related as
[6-2]
Effective focal length Actual focal length sin u

where u is the anode angle. Foreshortening of the focal spot length at the central ray
is called the line focus principle, as described by Equation 6-2. An ability to have a
smaller effective focal spot size for a large actual focal spot increases the power loadings for smaller effective focal spot sizes.
Example 1: The actual anode focal area for a 20-degree anode angle is 4 mm (length) by
1.2mm (width). What is the projected focal spot size at the central axis position?
Answer: Effective length = actual length sin = 4 mm sin 20 degrees = 4 mm 0.34 =
1.36mm; therefore, the projected focal spot size is 1.36 mm (length) by 1.2 mm (width).
Example 2: If the anode angle in Example 1 is reduced to 10 degrees and the actual focal spot
size remains the same, what is the projected focal spot size at the central axis position?
Answer: Effective length = 4 mm sin 10 degrees = 4 mm 0.174 = 0.69 mm; thus, the smaller
anode angle results in a projected size of 0.69 mm (length) by 1.2 mm (width) for the same actual
target area.
As the anode angle decreases (approaches 0 degrees), the effective focal spot
becomes smaller for the same actual focal area, providing better spatial resolution
of the object when there is geometric image magnification. Also, for larger actual
focal areas, greater x-ray output intensity with shorter exposure times is possible.
However, a small anode angle limits the usable x-ray size at a given source to image
receptor distance, because of cutoff of the beam on the anode side of the beam. Field
coverage is also less for short focus-to-detector distances (Fig. 6-14). Therefore, the
optimal anode angle depends on the clinical imaging application. A small anode
angle (~7 to 9 degrees) is desirable for small field-of-view devices, such as some small
fluoroscopy detectors, where field coverage is limited by the image receptor diameter
183

A
Large anode angle

Small filament length
Large anode angle

Long filament length
Small anode angle

Long filament length
Actual
Size
Actual
Size
Actual
Size
Projected
Size
Projected
Size
Large field coverage
Small effective focal spot
Poor power loading
Large field coverage

Large effective focal spot
Good power loading
Small field coverage

Small effective focal spot
Good power loading
FIGURE 6-14 Field coverage and effective focal spot length vary with the anode angle. A. A large anode
angle provides good field coverage at a given distance; however, to achieve a small effective focal spot, a
small actual focal area limits power loading. B. A large anode angle provides good field coverage, and achievement of high power loading requires a large focal area; however, geometric blurring and image degradation
occur. C. A small anode angle limits field coverage at a given distance; however, a small effective focal spot is
achieved with a large focal area for high power loading.
(e.g., 23 cm). Larger anode angles (~12 to 15 degrees) are necessary for general
radiographic imaging to achieve sufficiently large field area coverage at typical focal
spot-to-detector distances such as 100 cm.
The effective focal spot length varies with the position in the image plane, in the
anode-cathode (AC) direction. Toward the anode side of the field, the projected
length of the focal spot shortens, whereas it lengthens towards the cathode side of
the field (Fig. 6-15). The width of the focal spot does not change appreciably with
position in the image plane. Nominal focal spot size (width and length) is specified
at the central ray of the beam, from the focal spot to the image receptor, perpendicular to the anode-cathode axis and bisecting the plane of the image receptor.
x-Ray mammography is an exception, where half-field geometry is employed, as
explained in Chapter 8.
Measurement and verification of focal spot size can be performed in several ways.
Common tools for measuring focal spot size are the pinhole camera, slit camera, star pattern, and resolution bar pattern (Fig. 6-16). The pinhole camera uses a very small circular
aperture (10 to 30 mm diameter) in a thin, highly attenuating metal (e.g., lead, tungsten,
or gold) disk to project a magnified image of the focal spot onto an image receptor. With
the pinhole camera positioned on the central axis between the x-ray source and detector,
an image of the focal spot is recorded. Figure 6-16E shows magnified (2) pinhole pictures of the large (top row) and small (bottom row) focal spots with a typical bi-gaussian
intensity distribution. Correcting for the known image magnification allows measurement of the focal spot dimensions. The slit camera consists of a highly attenuating metal
(usually tungsten) plate with a thin slit, typically 10 mm wide. In use, the slit camera is
positioned above the image receptor, with the center of the slit on the central axis, and
with the slit either parallel or perpendicular to the A-C axis. Measuring the width of the
x-ray distribution in the image and correcting for magnification yields one dimension of
184
Actual focal spot
Width
Projected
focal spot
Focal spot size variation in the field

Anode
15
Length
Anode
Length
10
5
0
5
Target Angle 15
10
Width
15
Cathode
Cathode
15 10 5 0
5 10 15
Degrees from central axis

FIGURE 6-15 Variation of the effective focal spot size in the image field occurs along the anode-cathode
direction. Focal spot distributions are plotted as a function of projection angle in degrees from the central axis,
the parallel (vertical axis), and the perpendicular (horizontal axis).
the focal spot. A second radiograph, taken with the slit perpendicular to the first, yields
the other dimension of the focal spot, as shown in Figure 6-16F. The star pattern test tool
(Fig. 6-16G) contains a radial pattern of lead spokes of diminishing width and spacing
on a thin plastic disk. Imaging the star pattern at a known magnification and measuring
the distance between the outermost blur patterns (location of the outermost unresolved
spokes as shown by the arrows) on the image allows the calculation of the effective
focal spot dimensions in the directions perpendicular and parallel to the A-C axis.
A large focal spot will have a greater blur diameter than a small focal spot, as shown
in the figure. A resolution bar pattern is a simple tool for evaluation of focal spot size
(Fig.6-16H). Bar pattern images demonstrate the effective resolution parallel and perpendicular to the A-C axis for a given magnification geometry, determined from the
number of the bar pattern that can be resolved.
The National Electrical Manufacturers Association (NEMA) has published
tolerances for measured focal spot sizes. For focal spot nominal (indicated) sizes less
than 0.8 mm, the measured focal spot size can be larger by 50% (e.g., for a 0.6-mm
focal spot, the measured size can be up to 0.9 mm), but not smaller than the nominal size. For focal spots between 0.8 and 1.5 mm nominal size, the measured focal
spot size can be 0% smaller to 40% larger; and for focal spots greater than 1.5 mm,
0%smaller to 30% larger.
Focal spot blooming is an increase in the size of the focal spot resulting from
high tube current (mA), and is caused by electron repulsion in the electron beam
between the cathode and anode. It is most pronounced at low kVs. Focal spot thinning is a slight decrease in the measured size with increasing kV (electron repulsion
and spreading in the x-ray tube is reduced). NEMA standards require measurement
at 75 kV using 50% of the maximal rated mA for each focal spot.
Heel Effect
The heel effect refers to a reduction in the x-ray beam intensity toward the anode side
of the x-ray field (Figure 6-17), caused by the greater attenuation of x-rays directed
toward the anode side of the field by the anode itself. The heel effect is less prominent with a longer source-to-image distance (SID). Since the x-ray beam intensity is
greater on the cathode side of the field, the orientation of the x-ray tube cathode over
thicker parts of the patient can result in a better balance of x-ray photons transmitted

A
Pinhole
Slit
Star
Bar
185
FIGURE 6-16 Various tools allow measurement of the focal spot size, either directly or indirectly. A and E:
Pinhole camera and images. B and F: Slit camera and images. C and G: Star pattern and images. D and H: Resolution bar pattern and images. For EH, the top row of images represents the measurements of the large focal spot
(1.2 mm 1.2 mm), and the bottom row the small focal spot (0.6 mm 0.6 mm). The star and bar patterns
provide an equivalent focal spot dimension based upon the resolvability of the equivalent spatial frequencies.
through the patient and onto the image receptor. For example, the preferred orientation of the x-ray tube for a chest x-ray of a standing patient would be with the A-C
axis vertical, and the cathode end of the x-ray tube down.
Off-Focal Radiation
Off-focal radiation results from electrons that scatter from the anode, and are
re-accelerated back to the anode, outside of the focal spot area. These electrons cause
low-intensity x-ray emission over the entire face of the anode, as shown in Figure6-18,
increasing patient exposure, causing geometric blurring, reducing image contrast, and
increasing random noise. A small lead collimator aperture placed near the x-ray tube
output port can reduce off-focal radiation by intercepting x-rays that are produced
away from the focal spot. An x-ray tube that has a metal enclosure and the anode at
electrical ground potential will have less off-focal radiation, because many of the scattered electrons are attracted to the metal envelope instead of the anode.
186
FIGURE 6-17 The heel effect is a loss of intensity on the anode side of the x-ray field of view. It is caused by
attenuation of the x-ray beam by the anode. Upper right is an expanded view that shows electrons interacting at depth within the anode and the resultant self attenuation of produced x-rays that have a trajectory
towards the anode side of the field.
x-ray Tube Insert

The x-ray tube insert contains the cathode, anode, rotor assembly, and support structures sealed in a glass or metal enclosure under a high vacuum. The high vacuum
prevents electrons from colliding with gas molecules and is necessary in most
electron beam devices. As x-ray tubes age, trapped gas molecules percolate from
FIGURE 6-18 Off-focal radiation is produced from
back-scattered electrons that are re-accelerated to the
anode outside the focal spot. This causes a low-intensity,
widespread radiation distribution pattern. Hotspots outside the focal spot indicate areas where the electrons
are more likely to interact.
Pinhole
Off-focal radiation distribution
187
tube structures and degrade the vacuum. A getter circuit is used to trap gas in the
insert and to maintain the vacuum.
x-rays are emitted in all directions from the focal spot; however, the x-rays that
emerge through the tube port constitute the useful beam. Except for mammography
and special-purpose x-ray tubes, the port is typically made of the same material as
the tube enclosure. Mammography tubes use beryllium (Z = 4) in the port to minimize absorption of the low-energy x-rays used in mammography.
x-ray Tube Housing

The x-ray tube housing supports, insulates, and protects the x-ray tube insert from
the environment. Between the x-ray tube insert and housing is oil that provides heat
conduction and electrical insulation. In many radiographic x-ray tubes, an expanding bellows inside the housing accommodates oil expansion due to heat absorption
during operation. If the oil heats excessively, a microswitch disables the operation
of the x-ray tube until sufficient cooling has occurred. x-ray tubes used in interventional fluoroscopy and CT commonly have heat exchangers to allow prolonged
operation at high output.
Lead shielding inside the housing attenuates nearly all x-rays that are not directed
to the tube port (see Fig. 6-7 for the typical lead sheet thickness and location within
the housing). A small fraction of these x-rays, known as leakage radiation, penetrates
the housing. Federal regulations (21 CFR 1020.30) require manufacturers to provide sufficient shielding to limit the leakage radiation exposure rate to 0.88 mGy air
kerma per hour (equivalent to 100 mR/h) at 1 m from the focal spot when the x-ray
tube is operated at the leakage technique factors for the x-ray tube. Leakage techniques are the maximal operable kV (kVmax, typically 125 to 150 kV) at the highest
possible continuous current (typically 3 to 5 mA at kVmax for most diagnostic tubes).
Each x-ray tube housing assembly has a maximal rated tube potential that must not
be exceeded during clinical operation of the x-ray tube source assembly. The x-ray
equipment is designed to prevent the selection of x-ray tube kV greater than the
maximal rating.
Collimators
Collimators adjust the size and shape of the x-ray field emerging from the tube port.
The collimator assembly typically is attached to the tube housing at the tube port
with a swivel joint. Two pairs of adjustable parallel-opposed lead shutters define a
rectangular x-ray field (Fig. 6-19). In the collimator housing, a beam of light reflected
by a mirror of low x-ray attenuation mimics the x-ray beam. Thus, the collimation of
the x-ray field is identified by the collimators shadows. Federal regulations (21 CFR
1020.31) require that the light field and x-ray field be aligned so that the sum of the
misalignments, along either the length or the width of the field, is within 2% of the
SID. For example, at a typical SID of 100 cm (40 inches), the sum of the misalignments between the light field and the x-ray field at the left and right edges must not
exceed 2 cm, and the sum of the misalignments at the other two edges also must not
exceed 2 cm.
Positive beam limitation (PBL) collimators automatically limit the field size
to the useful area of the detector. Mechanical sensors in the film cassette holder
detect the cassette size and location and automatically adjust the collimator
blades so that the x-ray field matches the cassette dimensions. Adjustment to a
smaller field area is possible; however, a larger field area requires disabling the
PBL circuit.
188
X-ray tube collimation

Central axis view
Mirror
Collimator
light
Collimator
Assembly
Collimator Blades
X-ray and light fields
FIGURE 6-19 The x-ray tube collimator assembly is attached to the housing at the tube port, typically on a
collar that allows it to be rotated. A light source, positioned at a virtual focal spot location, illuminates the field
from a 45-degree angle mirror. Lead collimator blades define both the x-ray and light fields.
Special x-ray Tube Designs

A grid-biased tube has a focusing cup that is electrically isolated from the cathode
filament and maintained at a more negative voltage. When the bias voltage is sufficiently large, the resulting electric field stops the tube current. Turning off the grid
bias allows the tube current to flow and x-rays to be produced. Grid biasing requires
approximately 4,000 V applied to the focusing cup with respect to the filament
to switch the x-ray tube current off (see Fig. 6-10). The grid-biased tube is used in
applications such as pulsed fluoroscopy and cine-angiography, where rapid x-ray
pulsing is necessary. Biased x-ray tubes are significantly more expensive than conventional, nonbiased tubes.
Mammography tubes are designed to provide the low-energy x-rays necessary to
produce optimal mammographic images. As explained in Chapter 8, the main differences between a dedicated mammography tube and a conventional x-ray tube are the
target materials (molybdenum, rhodium, and tungsten), the output port (beryllium
versus glass or metal insert material), the smaller effective focal spot sizes (typically
0.3 and 0.1 mm nominal focal spot sizes), and the use of grounded anodes.
x-ray tubes for interventional fluoroscopy and CT require high instantaneous x-ray
output and high heat loading and rapid cooling. Furthermore, in CT, with the fast x-ray
tube rotation (as low as 0.3 s for a complete rotation about the patient) and the tremendous mechanical forces it places on the CT tube, planar surface cathode emitter designs
different than the common helical filaments and enhanced bearing support for the rotating anode are often used. One manufacturers CT tube incorporates a novel design with
the cathode and the anode as part of a metal vacuum enclosure that is attached to externally mounted bearings and drive motor to rotate the assembly as shown in Figure 6-20.
Dynamic steering of the electron beam within the tube is achieved by external electromagnetic deflection coils to direct the electrons to distinct focal spots on the anode,
which can produce slightly different projections and improve data sampling during
the CT acquisition (refer to Chapter 10 on CT). Direct anode cooling by oil circulating
within the housing provides extremely high cooling rates, and eliminates the need for
high anode heat storage capacity. The cessation of imaging during clinical examinations
to allow anode cooling is seldom necessary when using these advanced x-ray tubes.

Deflection
Oil heat exchange/cooling
coils
Ceramic
Insulator
Vacuum
Bearings
Rotation
Motor
Cathode
Electron beams
X-ray beams
189
FIGURE 6-20 Diagram of an advanced CT x-ray

tube, showing the anode and the planar cathode
within a rotating vacuum enclosure. The bearings are
mounted outside of the vacuum enclosure. Deflection coils magnetically direct the electron beam to
specific areas on the target. Circulating oil rapidly
removes excess heat from the anode. The electron
beam can be rapidly deflected between two focal
spots; this is known as a flying focal spot.
Tube housing
Anode
Recommendations to Maximize x-ray Tube Life

x-ray tubes eventually must be replaced, but a long lifetime can be achieved with
appropriate care and use. Several simple rules are discussed here. (1) Minimize filament boost prep time (the first detent of two on the x-ray exposure switch) especially
when high mA is used. If applied for too long, filament life will be shortened, unstable
operation will occur, and evaporated tungsten will be deposited on the glass envelope. (2) Use lower tube current with longer exposure times to arrive at the desired
mAs if possible. (3) Avoid extended or repeated operation of the x-ray tube with high
technique (kV and mAs) factors because, even though the x-ray generator has logic to
prohibit single exposure settings that could damage the x-ray tube, multiple exposures
could etch the focal track, resulting in less radiation output; transmit excessive heat to
the bearings; and cause outgassing of the anode structure that will cause the tube to
become unstable. (4) Always follow the manufacturers recommended warm-up procedure. Do not make high mA exposures on a cold anode, because uneven expansion
caused by thermal stress can cause cracks. (5) Limit rotor start and stop operations,
which can generate significant heating and hot spots within the stator windings; when
possible, a 30 to 40 s delay between exposures should be used.
Filtration
As mentioned earlier, filtration is the removal of x-rays as the beam passes through
a layer of material. Filtration includes both the inherent filtration of the x-ray tube
and added filtration. Inherent filtration includes the thickness (1 to 2 mm) of the
glass or metal insert at the x-ray tube port. Glass (primarily silicon dioxide, SiO2) and
aluminum have similar attenuation properties (ZSi = 14 and ZAl = 13) and effectively
attenuate all x-rays in the spectrum below about 15 keV. Dedicated mammography
tubes, on the other hand, require beryllium (Z = 4) to permit the transmission of lowenergy x-rays. Inherent filtration includes attenuation by housing oil and the field
light mirror in the collimator assembly.
Added filtration refers to sheets of metal intentionally placed in the beam to
change its effective energy. In general diagnostic radiology, added filters attenuate the low-energy x-rays in the spectrum that have almost no chance of penetrating the patient and reaching the x-ray detector. Because the low-energy x-rays are
absorbed by the filters instead of the patient, radiation dose is reduced by beam
filtration. Aluminum (Al) is the most commonly used added filter material. Other
common filter materials include copper and plastic (e.g., acrylic). An example of the
patient dose savings obtained with extra tube filtration is described in Section 6.5,
190
point no. 4, beam filtration. In mammography, thin filters of Mo, Rh, and silver (Ag)
are used to transmit bremsstrahlung x-rays in the intermediate energy range (15
to 25 keV), including characteristic radiation from Mo and Rh, and also to highly
attenuate lowest and highest x-ray energies in the spectrum (see Chapter 8). Some
have advocated using rare earth elements (K-absorption edges of 39 to 65 keV) such
as erbium in filters in radiography to reduce patient dose and improve image contrast
when contrast materials are used.
Compensation (equalization) filters are used to change the spatial pattern of
the x-ray intensity incident on the patient, so as to deliver a more uniform x-ray
exposure to the detector. For example, a trough filter used for chest radiography has
a centrally located vertical band of reduced thickness and consequently produces
greater x-ray fluence in the middle of the field. This filter compensates for the high
attenuation of the mediastinum and reduces the exposure latitude incident on the
image receptor. Wedge filters are useful for lateral projections in cervical-thoracic
spine imaging, where the incident fluence is increased to match the increased tissue
thickness encountered (e.g., to provide a low incident flux to the thin neck area and
a high incident flux to the thick shoulders). Bow-tie filters are used in CT to reduce
dose to the periphery of the patient, where x-ray paths are shorter and fewer x-rays
are required. Compensation filters are placed close to the x-ray tube port or just
external to the collimator assembly.
6.3 x-ray Generators

The principal function of an x-ray generator is to provide current at a high voltage to
an x-ray tube. Electrical power available to a hospital or clinic provides up to about
480 V, much lower than the 20,000 to 150,000 V needed for x-ray production.
Transformers are principal components of x-ray generators; they convert low voltage
into high voltage through a process called electromagnetic induction.
Electromagnetic Induction and Voltage Transformation

Electromagnetic induction is a phenomenon in which a changing magnetic field
induces an electrical potential difference (voltage) in a nearby conductor and also
in which a voltage is induced in a conductor moving through a stationary magnetic
field. For example, the changing magnetic field from a moving bar magnet induces
a voltage and a current in a nearby conducting wire (Fig. 6-21A). As the magnet
moves in the opposite direction away from the wire, the induced current flows in the
opposite direction. The magnitude of the induced voltage is proportional to the rate
of change of the magnetic field strength.
Electrical current, such as the electrons flowing through a wire, produces a magnetic field whose magnitude (strength) is proportional to the magnitude of the current (see Fig. 6-21B). For a coiled wire geometry, superimposition of the magnetic
fields from adjacent turns of the wire increases the amplitude of the overall magnetic field (the magnetic fields penetrate the wires insulation), and therefore the
magnetic field strength is proportional to the number of wire turns. A constant current flowing through a wire or a coil produces a constant magnetic field, and a varying current produces a varying magnetic field. With an AC and voltage, such as the
standard 60cycles/s (Hz) AC in North America and 50 Hz AC in most other areas of
the world, the induced magnetic field alternates with the current.
191
A Changing magnetic field induces electron flow:

Induced electron
flow in conductor
Forward motion
Reverse motion
Reversed
electron flow
B Current (electron flow) in a conductor creates a magnetic field; its

amplitude and direction determines magnetic field strength and polarity
N
N
N
N
Current:
Direction:
Large
forward
Small
forward
Small
reverse
Large
reverse
FIGURE 6-21 Principles of electromagnetic induction are illustrated. A. Induction of an electrical current in
a wire conductor coil by a moving (changing) magnetic field. The direction of the current is dependent on the
direction of the magnetic field motion. B. Creation of a magnetic field by the current in a conducting coil. The
polarity and magnetic field strength are determined by the amplitude and direction of the current.
A wire or a wire coil with a changing current will induce a voltage in a nearby
wire or wire coil, and therefore, an AC applied to a wire or a wire coil will induce
an alternating voltage in another wire or wire coil by electromagnetic induction.
However, when a constant current, like that produced by a chemical battery, flows
through a wire or a wire coil, although it creates a constant magnetic field, electromagnetic induction does not occur, and so it does not induce a voltage or current in
a nearby conductor.
Transformers
Transformers use the principle of electromagnetic induction to change the voltage
of an electrical power source. The generic transformer has two distinct, electrically
insulated wires wrapped about a common iron core (Fig. 6-22). Input AC power
produces an oscillating magnetic field on the primary winding of the transformer,
where each turn of the wire amplifies the magnetic field that is unaffected by the electrical insulation and permeates the iron core. Contained within the core, the changing magnetic field induces a voltage on the secondary winding, the magnitude of
which is amplified by the number of turns of wire. The voltage induced in the second
winding is proportional to the voltage on the primary winding and the ratio of the
number of turns in the two windings, as stated by the Law of Transformers,
VP N P
VS N S
[6-3]
192
FIGURE 6-22 The basic transformer

consists of an iron core, a primary winding
circuit, and a secondary winding circuit.
An AC flowing through the primary winding produces a changing magnetic field,
which permeates the core and induces
an alternating voltage on the secondary
winding. This mutual electromagnetic
induction is mediated by the containment
of the magnetic field in the iron core and
permeability through wire insulation.
Iron core: conduit for changing magnetic field

Primary
winding
e
Secondary
Winding
m
AC in
AC out
where Np is the number of turns in the primary coil, Ns is the number of turns in the
secondary coil, Vp is the amplitude of the input voltage on the primary side of the
transformer, and Vs is the amplitude of the output voltage on the secondary side.
A transformer can increase, decrease, or isolate input voltage, depending on the
ratio of the numbers of turns in the two coils. For Ns Np, a step-up transformer
increases the secondary voltage; for Ns Np, a step-down transformer decreases
the secondary voltage; and for Ns = Np, an isolation transformer produces a secondary
voltage equal to the primary voltage. Configurations of these transformers are shown
in Figure 6-23. An input AC waveform is supplied to a transformer in order to produce a changing magnetic field and induce a voltage in the secondary winding. A
step-up transformer circuit provides the high voltage necessary (20 to 150 kV) for a
diagnostic x-ray generator.
For electrons to be accelerated to the anode in an x-ray tube, the voltage at the
anode must be positive with respect to the cathode, but alternating waveforms change
between negative and positive voltage polarity each half cycle. For continuous production of x-rays, the anode must be continuously at a positive voltage with respect
to the cathode. However, this occurs only half of the time if an alternating voltage
waveform is provided directly to the x-ray tube. A basic electrical component known
as a rectifier will allow current to flow in one direction only. For instance, the x-ray
tube itself can act as a rectifier, since current usually will flow only when the anode
FIGURE6-23 Transformers
increase (step up), decrease (step
down), or leave unchanged (isolate)
the input voltage depending on the
ratio of primary to secondary turns,
according to the Law of Transformers.
In all cases, the input and the output
circuits are electrically isolated.
Primary
Input
Voltage
Secondary
Output
Voltage
Step-up transformer
Isolation transformer
Step-down transformer
193
has positive and the cathode has negative polarity; however, if the anode becomes
very hot from use and the accelerating voltage is applied with reverse polarity (tube
cathode positive and anode negative), electrons can be released from the hot anode
and accelerated into the cathode, possibly damaging it. A diode is a device with two
terminals. When a voltage is applied between the terminals with a specific polarity, there is very little resistance and a large current flows; when the same voltage is
applied with the opposite polarity, little or no current flows. Diodes come in all sizes,
from large, x-ray tubesized devices down to microscopic, solid-state components on
an integrated circuit board. Clever use of diodes arranged in a bridge rectifier circuit
can route the flow of electrons through an AC circuit to create a direct current (DC),
a unidirectional movement of electrons in which the voltage polarity never reverses.
Rectification is an important function of the x-ray generator.
Power is the rate of energy production or expenditure per unit time. The SI unit
of power is the watt (W), defined as 1 joule (J) of energy per second. For electrical
devices, power is equal to the product of voltage and current.

P=IV
[6-4]
Because a volt is defined as 1 joule per coulomb and an ampere is 1 coulomb per
second,
1 watt = 1 volt 1 ampere
For an ideal transformer, because the power output is equal to the power input,
the product of voltage and current in the primary circuit is equal to that in the secondary circuit

VPI P = VSI S
[6-5]
where IP is the input current on the primary side and IS is the output current on the
secondary side. Therefore, a decrease in current must accompany an increase in voltage, and vice versa. Equations 6-3 and 6-5 describe ideal transformer performance.
Power losses in an actual transformer due to inefficient coupling cause both the
voltage and current on the secondary side of the transformer to be less than those
predicted by these equations.
Example: The ratio of primary to secondary turns is 1:1,000 in a transformer. If an input AC
waveform has a peak voltage of 50 V, what is the peak voltage induced in the secondary side?
VP NP 50
1
;
;VS 50 1,000 50,000 V 50kV

VS NS VS 1000
What is the secondary current for a primary current of 10 A?
VP IP VS IS ; 50 V 10 A 50,000 V IS ; IS 0.00110 A 10 mA
The high-voltage section of an x-ray generator contains a step-up transformer, typically with a primary-to-secondary turns ratio of 1:500 to 1:1,000. Within this range,
a tube voltage of 100 kV requires an input line voltage of 200 to 100 V, respectively. The center of the secondary winding is usually connected to ground potential
(center tapped to ground). Ground potential is the electrical potential of the earth.
Center tapping to ground does not affect the maximum potential difference applied
194
between the anode and cathode of the x-ray tube, but it limits the maximum voltage
at any point in the circuit relative to ground to one half of the peak voltage applied
to the tube. Therefore, the maximum voltage at any point in the circuit for a centertapped transformer of 150 kV is 75 kV or +75 kV, relative to ground. This reduces
electrical insulation requirements and improves electrical safety. In some x-ray tube
designs (e.g., mammography and CT), the anode is maintained at the same potential as the body of the insert, which is maintained at ground potential. Even though
this places the cathode at peak negative voltage with respect to ground, the low kV
(less than 50 kV) used in mammography does not present a big electrical insulation
problem, while in modern CT systems (up to 140 kV) the x-ray generator is placed
adjacent to the x-ray tube in the enclosed gantry.
x-ray Generator Modules

Modules of the x-ray generator (Fig. 6-24) include the high-voltage power circuit,
the stator circuit, the filament circuit, the focal spot selector, and automatic exposure control (AEC) circuit. Generators typically have circuitry and microprocessors
that monitor the selection of potentially damaging overload conditions in order to
protect the x-ray tube. Combinations of kV, mA, and exposure time delivering excessive power to the anode are identified, and such exposures are prohibited. Heat
load monitors calculate the thermal loading on the x-ray tube anode, based on kV,
mA, and exposure time, and taking cooling into account. Some x-ray systems are
equipped with sensors that measure the temperature of the anode. These systems
protect the x-ray tube and housing from excessive heat buildup by prohibiting exposures that would damage them. This is particularly important for CT scanners and
high-powered interventional fluoroscopy systems.
Operator Console
For radiographic applications, a technologist at the operators console can select the
tube voltage (kV), the tube current (mA), the exposure time (s), or the product of
mA and time (mAs), the AEC mode, the AEC sensors to be used, and the focal spot.
Control desk
Microprocessors
Power circuit
Stator
circuit
+
-
kV selector
mAs selector
kV control
Fluoroscopy selector
AEC --ABC
mA and mAs
control
Filament
circuit
Interface controllers
Service
Phototimer
circuits
FIGURE 6-24 A modular schematic view shows the basic x-ray generator components. Most systems are
now microprocessor controlled and include service support diagnostics.
195
If AEC is used, exposure time is not set. The focal spot size (i.e., large or small)
is usually determined by the mA setting; low mA selections allow the small focal
spot to be used, and higher mA settings require the use of the large focal spot due
to anode heating concerns. On some x-ray generators, preprogrammed techniques
can be selected for various examinations (e.g., chest; kidneys, ureter and bladder;
cervical spine; and extremities). For fluoroscopic procedures, although kV and mA
can be manually selected, the generators automatic exposure rate control circuit,
sometimes called the automatic brightness control (ABC), is commonly activated. It
automatically sets the kV and mA from feedback signals from a sensor that indicates
the radiation intensity at the image receptor. All console circuits have relatively low
voltages and currents to minimize electrical hazards.
High-Frequency x-ray Generator

Several x-ray generator circuit designs are in use, including single-phase, three-phase,
constant potential, and high-frequency inverter generators. The high-frequency generator is now the contemporary state-of-the-art choice for diagnostic x-ray systems.
Its name describes its function, whereby a high-frequency alternating waveform (up
to 50,000 Hz) is used for efficient conversion of low to high voltage by a step-up
transformer. Subsequent rectification and voltage smoothing produce a nearly constant output voltage. These conversion steps are illustrated in Figure 6-25. The operational frequency of the generator is variable, depending on the exposure settings (kV,
mA, and time), the charge/discharge characteristics of the high-voltage capacitors on
the x-ray tube, and the frequency-to-voltage characteristics of the transformer.
Figure 6-26 shows the components and circuit diagram of a general-purpose high-frequency inverter generator. Low-frequency, low-voltage input power
(50 to 60cycles/s AC) is converted to a low voltage, direct current. Next, an inverter
circuit creates a high-frequency AC waveform, which supplies the high-voltage transformer to create a high-voltage, high-frequency waveform. Rectification and smoothing produces high-voltage DC power that charges the high-voltage capacitors placed
across the anode and cathode in the x-ray tube circuit. Accumulated charge in the
capacitors will produce a voltage to the x-ray tube according to the relationship V =
Q/C, where V is the voltage (volts), Q is the charge (coulombs), and C is the capacitance (farads). During the x-ray exposure, feedback circuits monitor the tube voltage
Single phase input voltage
+
Inputvoltage
Rectifier
Smooth
Inverter
110V60Hz
Transformer
AC 500-40000 Hz
Rectifier
Smooth
X-ray tube
Output voltage
50-150 kV
FIGURE 6-25 In a high-frequency

inverter generator, a single- or threephase AC input voltage is rectified
and smoothed to create a DC waveform. An inverter circuit produces a
high-frequency AC waveform as input
to the high-voltage transformer. Rectification and capacitance smoothing
provide the resultant high-voltage output waveform, with properties similar
to those of a three-phase system.
196
+
AC / DC
converter
High voltage
transformer and
rectifier circuits
DC / AC
inverter
DC
voltage
Ttigger
pulses
Single
phase
or
three
phase
input
power
+kVp/2
kVp/2
kVp
reference
Voltage
comparator
oscillator
mA
reference
Voltage
comparator
oscillator
kVp
sense
High voltage
capacitors
mA
sense
Trigger
pulses
Filament
Circuit
FIGURE 6-26 Modular components and circuits of the high-frequency generator. The selected high voltage
across the x-Ray tube is created by charging high voltage capacitors to the desired potential difference. During
the exposure when the x-ray circuit is energized, tube current is kept constant by the mA sense circuit that
maintains the proper filament current by sending trigger pulses to the filament circuit , and tube voltage is
kept constant by the kV sense circuit that sends trigger pulse signals to the DC/AC inverter to maintain the
charge of high voltage capacitors.
and tube current and continuously supply charge to the capacitors as needed to
maintain a nearly constant voltage.
For kV adjustment, a voltage comparator measures the difference between the
reference voltage (a calibrated value proportional to the requested kV) and the actual
kV measured across the tube by a voltage divider (the kV sense circuit). Trigger
pulses generated by the comparator circuit produce a frequency that is proportional
to the voltage difference between the reference signal and the measured signal.
Alarge discrepancy in the compared signals results in a high trigger-pulse frequency,
whereas no difference produces few or no trigger pulses. For each trigger pulse,
the DC/AC inverter circuit produces a corresponding output pulse, which is subsequently converted to a high-voltage output pulse by the transformer. The highvoltage capacitors store the charge and increase the potential difference across the
x-ray tube. When the requested x-ray tube voltage is reached, the output pulse
rate of the comparator circuit settles down to a constant value, and recharging of
the high-voltage capacitors is constant. When the actual tube voltage drops below
a predetermined limit, the pulse rate increases. The feedback pulse rate (generator
frequency) strongly depends on the tube current (mA), since the high-voltage capacitors discharge more rapidly with higher mA, thus actuating the kV comparator circuit. Because of the closed-loop voltage regulation, input line voltage compensation
is not necessary, unlike older generator designs.
The mA is regulated in an analogous manner to the kV, with a resistor circuit
sensing the actual mA (the voltage across a resistor is proportional to the current)
and comparing it with a reference voltage. If the mA is too low, the mA comparator
circuit increases the trigger frequency, which boosts the power to the filament to
raise its temperature and increase the thermionic emission of electrons. The feedback
197
circuit eliminates the need for space charge compensation circuits and automatically
corrects for filament aging effects.
There are several advantages to the high-frequency inverter generator. Single-phase
or three-phase input voltage can be used. Closed-loop feedback and regulation circuits
ensure reproducible and accurate kV and mA values. Transformers operating at high
frequencies are efficient, compact, and less costly to manufacture than those in other
generator designs such as single-phase, three-phase, or constant-potential generators.
Modular and compact design makes equipment installation and repairs relatively easy.
The high-frequency inverter generator is the preferred system for all but a few
applications (e.g., those requiring extremely high power, extremely fast kV switching, or submillisecond exposure times provided by a constant-potential generator,
which is very costly and requires more space).
Voltage Ripple
Ideally, the voltage applied to an x-ray tube would be constant. However, variations occur
in the high voltage produced by an x-ray generator and applied to the x-ray tube. In an
electronic waveform, voltage ripple is defined as the difference between the peak voltage
and the minimum voltage, divided by the peak voltage and m
ultiplied by 100%:
% voltage ripple

Vmax Vmin
100
Vmax

[6-6]
The voltage ripple for various types of x-ray generators is shown in Figure 6-27.
In theory, a single-phase generator, whether one-pulse or two-pulse output, has 100%
voltage ripple. Actual voltage ripple for a single-phase generator is less than 100%
because of cable capacitance effects (longer cables produce a greater capacitance,
whereby the capacitance borrows voltage from the cable and returns it a short time
later, smoothing the peaks and valleys of the voltage waveform). Three-phase 6-pulse
and 12-pulse generators (not discussed in this chapter) have voltage ripples of 3% to
25%. High-frequency generators have a ripple that is kV and mA dependent, typically
Generator
type
Typical voltage
waveform
kV
ripple
Single-phase 1-pulse
(self rectified)
100%
Single-phase 2-pulse
(full wave rectified)
100%
3-phase
6-pulse
13% - 25%
3-phase
12-pulse
3% - 10%
Mediumhigh frequency
inverter
4% - 15%
Constant
Potential
<2%
FIGURE 6-27 Typical voltage

ripple for various x-ray generators
used in diagnostic radiology varies
from 100% voltage ripple for a
single-phase generator to almost
no ripple for a constant-potential
generator.
198
similar to a three-phase generator, ranging from 4% to 15%. Higher kV and mA settings result in less voltage ripple for these generators. Constant-potential generators
have an extremely low ripple, less than 2%, but are expensive and bulky.
Timers
Digital timers have largely replaced electronic timers based on resistor-capacitor
circuits and charge-discharge timers in older systems. Digital timer circuits have
extremely high reproducibility and microsecond accuracy, but the precision and
accuracy of the x-ray exposure time depends chiefly on the type of switching
(high voltage, low voltage, or x-ray tube switching) employed in the x-ray system.
A countdown timer, also known as a backup timer, is used as a safety mechanism to
terminate the exposure in the event of an exposure switch or timer failure.
Switches
The high-frequency inverter generator typically uses electronic switching on the
primary side of the high-voltage transformer to initiate and stop the exposure.
Arelatively rapid response resulting from the high-frequency waveform characteristics
of the generator circuit allows exposure times as short as 2 ms.
Alternatively, a grid-controlled x-ray tube can be used with any type of generator
to switch the exposure on and off by applying a bias voltage (about ~4,000 V) to the
focusing cup. This is the fastest switching method, with minimal turn-onturnoff lag;
however, there are extra expenses and high-voltage insulation issues to be considered.
PhototimerAutomatic Exposure Control

The phototimer, also known as the automatic exposure control (AEC) system, is often
used instead of manual exposure time settings in radiography. Phototimers measure the
actual amount of radiation incident on the image receptor (i.e., screen-film or digital
radiography detector) and terminate x-ray production when the proper radiation exposure is obtained. Compensation for patient thickness and other variations in attenuation are achieved at the point of imaging. A phototimer system consists of one or more
radiation detectors, an amplifier, a film density or digital SNR variable selector, asignal
integrator circuit, a comparator circuit, a termination switch, and a backup timer safety
shutoff switch (Fig. 6-28). x-rays transmitted through the patient and antiscatter grid,
if present, generate ion pairs in one to three selectable ionization chambers positioned
prior to the detector. An amplifier boosts the signal, which is fed to a voltage comparator and integration circuit. When the accumulated signal equals a preselected reference value, an output pulse terminates the exposure. A user-selectable film density
or SNR selector on the generator control panel increases or decreases the reference
voltage about 10% to 15% per step to modify the total accumulated x-ray exposure.
For general diagnostic radiography, the phototimer sensors are placed in front of the
image receptor to measure the transmitted x-ray flux through the patient (see Fig. 6-28).
Positioning in front of the image receptor is possible because of the high transparency of
the ionization chambers at the high kV values (greater than 50 kV) used for diagnostic
radiography exams. In the event of a phototimer detector or circuit failure, the backup
timer safety switch terminates the x-ray exposure after a preset on time.
To facilitate radiography of various anatomic projections, wall-mounted chest and
table cassette stands and DR image receptors typically have three photocells arranged
as shown in Figure 6-28 (front view). The technologist can select which photocells
are used for each radiographic application. For instance, in posteroanterior chest
199

AEC (ion chamber) detectors
Side view
Front view
X-ray
generator
Terminate
Backup timer
4 3 21 0 +1+2 +3 +4
Low dose
Amplifier
High SNR
Reference Voltage
Accumulating Voltage
Digital Detector
Comparator
Signal to Noise control setting
FIGURE 6-28 AEC detectors measure the radiation incident on the detector and terminate the exposure
according to a preset optical density or signal-to-noise ratio achieved in the analog or digital image. A front view
(left) and side view (middle) of a chest cassette stand and the locations of ionization chamber detectors are
shown. The desired signal to the image receptor and thus the signal-to-noise ratio may be selected at the operators console with respect to a normalized reference voltage.
imaging, the two outside chambers are usually activated, and the x-ray beam transmitted through the lungs determines the exposure time. This prevents signal saturation in the lung areas, which can occur when the transmitted x-ray flux is otherwise
measured under the highly attenuating mediastinum with the center chamber.
6.4 Power Ratings and Heat Loading and Cooling

The power rating of an x-ray tube or generator is the maximal power that an x-ray
tube focal spot can accept or the generator can deliver. General diagnostic x-ray tubes
and x-ray generators use 100 kV and the maximum ampere rating available for a 0.1
s exposure as the benchmark, as

Power (kW) 100 kV I (A max for a 0.1 s exposure)
[6-7]
For instance, a generator that can deliver 800 mA (0.8 A maximum) of tube
current at 100 kV for 0.1 second exposure has a power rating of 80 kW according
to Equation 6-7. For applications such as computed tomography and interventional
fluoroscopy, higher- power x-ray generators are specified, typically of 80to 100 kW
or higher. However, this power capability must be matched to the power deposition
capability of the focal spot; otherwise, the power cannot be realized without exceeding the x-ray tube heat limitations. Focal spot dimensions of approximately 1.2 mm
1.2 mm, large diameter anodes, and fast rotation speeds are necessary to achieve 80
to 100 kW of power deposition. Most medium focal spot dimensions (0.6 mm 0.6
mm) have moderate power ratings (30 to 50 kW), and smaller focal spots (0.3mm
0.3 mm) have low power ratings (5 to 15 kW), but power ratings are dependent
on several engineering, operational, and x-ray tube design factors. For instance, in
modern CT x-ray tubes, a shallow anode angle (7), bearings mounted outside of the
vacuum of the insert, and heat exchangers allow for extended high power operation
200
(e.g., continuous operation at 800 mA at 120 kV for minutes). x-ray generators have
circuits or programs to prohibit combinations of kV, mA, and exposure time that will
exceed the single-exposure power deposition tolerance of the focal spot.
x-ray generator power ratings vary considerably. The highest generator power
ratings (80 to 120 kW) are found in interventional radiology and cardiovascular
fluoroscopic imaging suites, while modern multirow detector CT scanners have 85to 100-kW generators. General radiographic or radiographic/fluoroscopic systems
use generators providing 30 to 80 kW. Lower powered generators (5 to 30 kW) are
found in mobile radiography and fluoroscopy systems, dental x-ray systems, and
other systems that have fixed anode x-ray tubes. When purchasing an x-ray generator and x-ray tube system, the clinical application must be considered, as well as
the matching of the power ratings of the x-ray tube large focal spot and the x-ray
generator. Otherwise, there will be a mismatch in terms of capability, and most likely
needless expense.
Heat Loading
The Heat Unit
The heat unit (HU) is a traditional unit that provides a simple way of expressing the
energy deposition on and dissipation from the anode of an x-ray tube. The number of
HU can be calculated from the parameters defining the radiographic technique

Energy (HU) peak voltage (kV) tube current (mA) exposure time (s) [6-8]
Although Equation 6-8 is correct for single-phase generator waveforms, it

underestimates the energy deposition of three-phase, high-frequency, and constantpotential generators because of their lower voltage ripple and higher average voltage.
A multiplicative factor of 1.35 to 1.40 compensates for this difference, the latter value
being applied to constant-potential waveforms. For example, an exposure of 80 kV,
250 mA, and 100 ms results in the following HU deposition on the anode:
Single-phase generator: 80 250 0.100 = 2,000 HU
Three-phase or high-frequency inverter generator: 80 250
0.100 1.35 = 2,700 HU
For continuous x-ray production (fluoroscopy), the HU/s is defined as follows:
HU / s kV mA
[6-9]
Heat accumulates by energy deposition and simultaneously disperses by anode

cooling. The anode cools faster at higher temperatures, so that for most fluoroscopic
and long CT procedures, steady-state equilibrium exists and Equation 6-9 will overestimate the heat energy delivered to the anode. This is taken into consideration
by the anode thermal characteristics chart for heat input and heat output (cooling)
curves.
The Joule
The joule (J) is the SI unit of energy. One joule is deposited by a power of one watt
acting for 1 s (1 J = 1 W 1 s). The energy, in joules, deposited in the anode is calculated as follows:

Energy (J) Voltage (V) Tube current (A) Exposure time(s)
[6-10]
201
The root-mean-square voltage, VRMS, is the constant voltage that would deliver
the same power as a varying voltage waveform (e.g., for a single phase generator),
which would be substituted for V, and, from Equation 6-8 and 6-10, the relationship
between the deposited energy in joules and HU can be approximated as
Energy (HU) 1.4 Heat input (J)
[6-11]
Anode Heating and Cooling Chart

Multiple x-ray exposures, continuous x-ray tube operation with CT, and prolonged fluoroscopy in interventional procedures deliver significant heat energy
to the anode. An anode heating and cooling chart (Fig. 6-29) shows the anode heat
loading for various input powers (kW or HU/s) as the x-ray tube is operating, taking
into account the cooling that simultaneously occurs. For low-input power techniques, the heat loading increases to a plateau, where the cooling rate equalsthe
heating rate. At higher power inputs, heating exceeds cooling, and after a certain time, a limit is reached where x-ray production must be stopped and anode
cooling must be allowed prior to using the x-ray tube again. With larger anode
heat capacities, the anode cooling curve is steeper because higher temperatures
result in more rapid cooling (the black-body effect, where the radiative cooling
rate is proportional to T4, for absolute temperature T). The maximal anode heat
capacity is indicated by the peak value of the cooling curve on the y-axis (0 time)
of the chart. Each anode heating and cooling chart is specific to a particular x-ray
tube and housing assembly. Some high-powered x-ray tubes for CT exceed 8 MHU
anode heat loading with extremely high anode and x-ray tube housing cooling rates,
achieved through advanced anode, rotor bearing, and housing designs. An example
Anode Heating and Cooling Curve: CT Tube
6
28 kW
5
20 kW
Energy (MJ)
12 kW
8 kW
3
2 kW
Cooldown
1
0
0
10
15
20
35
25
30
Time (minutes)
40
45
50
55
60
FIGURE 6-29 Anode heating and cooling curve chart for a CT scanner plots energy in megajoules (MJ) on
the vertical axis and time in minutes on the horizontal axis. A series of power input curves from low (2 kW) to
high (28 kW) are determined by the kV and mA settings with continuous x-ray tube operation as a function
of time. The cooling curve shows the rate of cooling and indicates faster cooling with higher anode heat load
(temperature). In this example the maximum capacity is 5.7 MJ. For low power inputs, heating and cooling
rates eventually equilibrate and reach a steady state, as shown for the 2, 4, and 8 kW curves.
202
of a heating and cooling chart for a modern x-ray CT tube, with a maximum anode
heat capacity of 5.7 megajoules (MJ) (equivalent to 8.0 MHU), is shown in Figure
6-29. If the x-ray technique is high or a large number of sequential or continuous
exposures have been taken (e.g., during operation of a CT scanner), it can be necessaryto wait before reenergizing the x-ray tube to avoid damage to the anode. Use of
the specific anode cooling curve can determine how long to wait, based upon
the total amount of accumulated heat, the simultaneous cooling that occurs,
andthe amount of power that will be deposited during the next acquisition. Modern x-ray systems have sensors and protection circuits to prevent overheating.
6.5 Factors Affecting x-ray Emission

The output of an x-ray tube is often described by the terms quality, quantity, and
exposure. Quality describes the penetrability of an x-ray beam, with higher energy
x-ray photons having a larger half-value layer (HVL) and higher quality. (The HVL
is discussed in Chapter 3.) Quantity refers to the number of photons comprising the
beam. Exposure, defined in Chapter3, is nearly proportional to the energy fluence
of the x-ray beam. x-ray production efficiency, exposure, quality, and quantity are
determined by six major factors: x-ray tube target material, tube voltage, tube current, exposure time, beam filtration, and generator waveform.
1. Anode target material affects the efficiency of bremsstrahlung radiation production, with output exposure roughly proportional to atomic number. Incident
electrons are more likely to have radiative interactions in higher-Z materials
(see Equation 6-1). The energies of characteristic x-rays produced in the target
depend on the target material. Therefore, the target material affects the quantity of
bremsstrahlung photons and the quality of the characteristic radiation.
2. Tube voltage (kV) determines the maximum energy in the bremsstrahlung spectrum and affects the quality of the output spectrum. In addition, the efficiency
of x-ray production is directly related to tube voltage. Exposure is approximately
proportional to the square of the kV in the diagnostic energy range.

Exposure kV 2
[6-12]
For example, according to Equation 6-12, the relative exposure of a beam

generated with 80 kV compared with that of 60 kV for the same tube current
and exposure time is calculated as follows:
2
80
1.78
60
Therefore, the output exposure increases by approximately 78%
(Fig. 6-30). An increase in kV increases the efficiency of x-ray production and the
quantity and quality of the x-ray beam.
Changes in the kV must be compensated by corresponding changes in mAs
to maintain the same exposure. At 80 kV, 1.78 units of exposure occur for every
1unit of exposure at 60 kV. To achieve the original 1 unit of exposure, the mAs
must be adjusted to 1/1.78 = 0.56 times the original mAs, which is a reduction
of 44%. An additional consideration of technique adjustment concerns the x-ray
attenuation characteristics of the patient. To achieve equal transmitted exposure
through a typical patient (e.g., 20-cm tissue), the compensatory mAs varies with
approximately the fifth power of the kV ratio
FIGURE 6-30 x-Ray tube output

intensity varies as the square of tube
voltage (kV). In this example, the
same tube current and exposure
times (mAs) are compared for 60 to
120 kV. The relative area under each
spectrum roughly follows a squared
dependence (characteristic radiation
is ignored).
X-ray output spectrum

(Equal tube current & exposure time)
100000
Relative photon fluence
203
80000
60000
40000
20000
0
0
20
40
60
80
Energy (keV)
100
120
kVp1
mAs1 = mAs 2
kVp2

[6-13]
According to Equation 6-13, if a 60-kV exposure requires 40 mAs for a

proper exposure through a typical adult patient, at 80 kV the adjusted mAs is
approximately
5
60
40 mAs 9.5 mAs
80
or about one fourth of the original mAs. The value of the exponent (between four
and five) depends on the thickness and attenuation characteristics of the patient.
3. Tube current (mA) is proportional to the number of electrons flowing from
the cathode to the anode per unit time. The exposure of the beam for a given
kV and filtration is proportional to the tube current. Also the exposure time
is the duration of x-ray production. The quantity of x-rays is directly proportional to the product of tube current and exposure time (mAs), as shown in
Figure 6-31.
FIGURE 6-31 x-ray tube output
intensity is proportional to the mAs
(tube current and exposure time).
Shown is the result of increasing the
mAs by a factor of two and three,
with a proportional change in the
number of x-rays produced.
X-ray output spectrum at 80 kVp

(1, 2 and 3 times increased mAs)
100000
80000
60000
40000
20000
0
20
40
60
80
Energy (keV)
100
120
FIGURE 6-32 x-ray tube output intensity decreases and spectral quality (effective
energy) increases with increasing thickness
of added tube filters. Shown are spectra
with added filtration at the same kV and
mAs.
Effect of Added Tube Filters:

Spectral shift to higher effective energy
Relative Output
204
Fixed kV and mAs
Energy (keV)
4. Beam filtration modifies the quantity and quality of the x-ray beam by preferentially removing the low-energy photons in the spectrum. This reduces the
number of photons (quantity) and increases the average energy, also increasing
the quality (Fig. 6-32). For highly filtered beams, the mAs required to achieve
a particular x-ray intensity will be much higher than for lightly filtered beams;
therefore, it is necessary to know the HVL of the beam and the normalized x-ray
tube intensity per mAs (in addition to the geometry) to calculate the incident
exposure and radiation dose to the patient. One cannot simply use kV and mAs
for determining the proper technique or to estimate the dose to the patient
without this information. In the United States, x-ray system manufacturers must
TABLE 6-3MINIMUM HVL REQUIREMENTS FOR X-RAY
SYSTEMS IN THE UNITED STATES (21 CFR

1020.30)
DESIGNED
OPERATING RANGE
MEASURED x-RAY
TUBE VOLTAGE (kV)
MINIMUM HVL
(mm of aluminum)
51 kV
30
0.3
40
0.4
5170 kV
70 kV
50
0.5
51
1.3
60
1.5
70
1.8
71
2.5
80
2.9
90
3.2
100
3.6
110
3.9
120
4.3
130
4.7
140
5.0
150
5.4
Note: This table refers to systems manufactured after June 2006. It does not
include values for dental or mammography systems.

Effect of tube filtration on surface dose (microgray)
for same detector signal to noise ratio100 kVp
Surface radiation dose (Gy)
6000.0
0 mm Al
2.0 mm Al
0.1 mm Cu
0.2 mm Cu
5000.0
4000.0
3000.0
2000.0
80 kVp
60 kVp
205
FIGURE 6-33 Added x-ray tube filters

can significantly reduce patient dose.
Compared are measured entrance doses
(air kerma with backscatter) to 10, 20, and
30 cm sheets of PMMA when using phototimed exposures at 60, 80, and 100 kV,
respectively. The top curve represents the
nominal beam condition, then 2.0 mm Al,
0.1 mm Cu + 1 mm Al, and 0.2 mm Cu +
1 mm Al for the lowest curve. For constant
signal-to-noise ratio, the mAs is increased
to compensate for added filter attenuation,
as listed in Table 6-4.
1000.0
0.0
15
10
20
30
25
Attenuator thickness (cm)
comply with minimum HVL requirements specified in the Code of Federal Regulations, 21CFR 1020.30 (Table 6-3), which have been adopted by many state
regulatory agencies. A pertinent clinical example of the radiation dose savings
achievable with added filtration compared to the minimum filtration is shown
in Figure 6-33, comparing the surface entrance radiation dose to the same signal
generated in the detector for 0 mm Al (minimum filtration), 2 mm Al, 0.1 mm
Cu, and 0.2 mm Cu. With more filtration, the dose savings become greater. For
instance, as listed in Table 6-4, the comparison of 30 cm polymethylmethacrylate
(PMMA) at 100 kV indicates a dose savings of 49% when a 0.2 mm Cu filter is
added to the beam. To achieve this requires an increase in mAs of 37%, from 14.5
to 19.8 mAs at 100 kV. Even though the acquisition technique is much higher, the
entrance dose is much lower, without any loss of image quality, particularly for
digital imaging devices where image contrast and brightness enhancements are
easily applied. Using added filtration in a modern x-ray collimator assembly is as
TABLE 6-4TUBE FILTRATION, MEASURED CHANGES IN REQUIRED mAs,
AND MEASURED SURFACE DOSE (mGy) FOR EQUIVALENT

SIGNAL IN THE OUTPUT DIGITAL IMAGE
10 cm PMMA (60 kV)
20 cm PMMA (80 kV)
30 cm PMMA (100 kV)
TUBE
CURRENT
TUBE
CURRENT
TUBE
CURRENT
DOSE
(Gy)
%D
DOSE
(Gy)
DOSE
(Gy)
FILTRATION
mAs
%D
DOSE
%D
mAs
DOSE
%D
mAs
%D
DOSE
%D
0 mm Al
3.8
264
6.8
1,153
14.5
4,827
2 mm Al
5.0
32
188
29
8.2
21
839
27
16.5
14
3,613
25
0.1 mm
Cu + 1 mm Al
6.2
63
150
43
9.3
37
680
41
17.6
21
2,960
39
0.2 mm
Cu + 1 mm Al
8.8
132
123
53
11.2
65
557
52
19.8
37
2,459
49
Note: % D indicates the percentage change from the no added filtration measurement. For mAs, there is
an increase in the mAs required to compensate for increased attenuation of the beam by the added filters.
For surface dose, there is a decrease in the percentage surface dose change with more added filters for equivalent
absorbed signal in the detector because of less attenuation of the beam in the object.
206
100 % Ripple
5 % Ripple
20
40
60
80
100
Energy (keV)
FIGURE 6-34 Output intensity (bremsstrahlung) spectra for the same tube voltage (kV) and the same tube
current and exposure time (mAs) demonstrate the higher effective energy and greater output of a three-phase
or high-frequency generator voltage waveform (~5% voltage ripple), compared with a single-phase generator
waveform (100% voltage ripple).
simple as pushing the filter selection button, or automatically selecting the filter
with anatomical programming.
5. Generator waveform affects the quality of the emitted x-ray spectrum. For
the same kV, a single-phase generator provides a lower average potential difference than does a three-phase or high-frequency generator. Both the quality and
quantity of the x-ray spectrum are affected (Fig. 6-34).
The x-ray quantity is approximately proportional to Ztarget kV2 mAs. The
x-ray quality depends on the kV, the generator waveform, and the tube filtration.
Exposure depends on both the quantity and quality of the x-ray beam. Compensation for changes in kV with radiographic techniques requires adjustments of mAs on
the order of the fourth to fifth power of the kV ratio, because kV determines quantity,
quality, and transmission through the object, whereas mAs determines quantity only.
Added filters to the x-ray tube can significantly lower patient dose by selectively
attenuating the low energy x-rays in the bremsstrahlung spectrum, but require a
compensatory increase in the mAs.
In summary, x-rays are the basic radiologic tool for a majority of medical diagnostic imaging procedures. Knowledge of x-ray production, x-ray generators, and
x-ray beam control is important for further understanding of the image formation
process and the need to obtain the highest image quality at the lowest possible radiation exposure.
SUGGESTED READING
Krestel E, ed. Imaging systems for medical diagnostics. Berlin/Munich, Germany: Siemens Aktiengesellschaft,
1990.
McCollough CH. The AAPM/RSNA physics tutorial for residents: x-ray production. Radiographics
1997;17:967984.
Seibert JA. The AAPM/RSNA physics tutorial for residents: x-ray generators. Radiographics 1997;17:
15331557.
Chapter
Radiography
7.1 Geometry of Projection Radiography
Radiography is the production of a two-dimensional image from a three-dimensional object, the patient (Fig. 7-1). The procedure projects the x-ray shadows of
the patients anatomy onto the detector and is often called projection radiography.
The source of radiation in the x-ray tube is a small spot, and x-rays that are produced in the x-ray tube diverge as they travel away from this spot. Because of beam
divergence, the collimated x-ray beam becomes larger in area and less intense with
increasing distance from the source. Consequently, x-ray radiography results in some
magnification of the object being radiographed. Radiography is performed with the
x-ray source on one side of the patient, and the detector is positioned on the other
side of the patient. During the exposure, incident x-rays are differentially attenuated
by anatomical structures in the patient. A small fraction of the x-ray beam passes
through the patient unattenuated and is recorded on the radiographic detector, forming the latent radiographic image.
The geometry of projection transmission imaging is described in Figure 7-2.
Magnification can be defined simply as

L image
Lobject
[7-1]
where Limage is the length of the object as seen on the image and Lobject is the length of
the actual object. Due to similar triangles, the object magnification can be computed
using the source to object distance (a) and the object to detector distance (b)
a b

[7-2]
a
The magnification will always be greater than 1.0 but approaches 1.0 when a relatively flat object (such as a hand in radiography) is positioned in contact with the
detector, where b < 0. The magnification factor changes slightly for each plane perpendicular to the x-ray beam axis in the patient, and thus different anatomical structures are magnified differently. This characteristic means that, especially for thicker
body parts, an AP (anterior-posterior) image of a patient will have a subtly different
appearance to an experienced viewer than a PA (posterior-anterior) image of the
same patient. Knowledge of the magnification in radiography is sometimes required
clinically, for example, in angioplasty procedures where the diameter of the vessel
must be measured to select the correct size of a stent or angiographic balloon catheter. In these situations, an object of known dimension (e.g., a catheter with notches
at known spacing) is placed near the anatomy of interest, from which the magnification can be determined and a correction factor calculated.
Radiographic detectors, both screen-film and digital, should be exposed to an
x-ray intensity within a relatively small range. X-ray technique factors such as the
applied tube voltage (kV), tube current (mA), and exposure time will be discussed

207
208
ex
FIGURE 7-1 The basic geometry of radiographic imaging is illustrated. The patient is positioned between
the x-ray tube and the detector system (A), and the radiograph is acquired. The photocell (seen between the
grid and detector) determines the correct exposure levels to the detector (phototiming). In transmission imaging (B), the x-rays pass through each point in the patient, and the absorption of all the structures along each
ray path combine to produce the primary beam intensity at that location on the detector. The density at each
point in the image (C) reflects the degree of x-ray beam attenuation across the image. For example, the gray
scale under the red dot on (C) corresponds to the total beam attenuation along the line shown in (B).
later, but an important parameter that should be kept consistent is the distance
between the x-ray source and the detector. For most radiographic examinations, the
source to imager distance (SID) is fixed at 100 cm, and there are usually detents
in the radiographic equipment that help the technologists to set this distance. An
exception to the 100-cm (40 inches) SID is for upright chest radiography, where the
SID is typically set to 183 cm (72 inches). The larger SID used for chest radiography
reduces the differential magnification in the lung parenchyma.
The focal spot in the x-ray tube is very small but is not truly a point source,
resulting in magnification-dependent resolution loss. The blurring from a finite x-ray
source is dependent upon the geometry of the exam, as shown in Figure 7-3. The
length of the edge gradient (Lg) is related to the length of the focal spot (Lf) by
b
Lg L f a ,
[7-3]
FIGURE 7-2 The geometry of beam divergence is shown. The object

is positioned a distance a from the x-ray source, and the detector is positioned a distance (a b) from the source. From the principle of similar
triangles,
Limage
a+b
Lobject
Lobject
b
Limage
Chapter 7 Radiography
209
Lf
Lg
FIGURE 7-3 The magnification of a sharp edge in the patient will result in some blurring of that structure
because of the use of a focal spot that is not truly a point source. For a focal spot with width Lf, the intensity
distribution across this distributed (nonpoint) source will result in the edge being projected onto the image
plane. The edge will no longer be perfectly sharp, but rather its shadow will reflect the source distribution.
Most sources are approximately gaussian in shape, and thus the blurred profile of the edge will also appear
gaussian. The length of the blur in the image is related to the width of the focal spot by L b L .
g
a f
where b represents the magnification of the focal spot. The edge gradient is
a
measured using a highly magnified metal foil with a sharp edge. In most circumstances, higher object magnification increases the width of the edge gradient
and reduces the spatial resolution of the image. Consequently, in most cases, the
patient should be positioned as close as possible to the detector to reduce magnification. For thin objects that are placed in contact with the detector, the magnification is approximately 1.0, and there will be negligible blurring caused by the
finite dimensions of the focal spot. In some settings (e.g., mammography), a very
small focal spot is used intentionally with magnification. In this case, the small
focal spot produces much less magnification-dependent blur, and the projected
image is larger as it strikes the detector. This intentional application of magnification radiography is used to overcome resolution limitations of the detector, and
increases spatial resolution.
7.2 Screen-Film Radiography

Screen-film radiography defined the field of radiology for most of the 20th century. As other digital modalities such as CT and MRI became routinely used, picture archiving and communication systems (PACS) evolved. Radiography and then
mammography were the last modalities to become assimilated into the digital arena,
due to the large display monitors and storage requirements for large-format, highresolution digital radiographic images. Nevertheless, screen-film radiography still is
used in low-volume clinics and other niche areas.
210

A
Screen film casette

Top view of TGrains
Top intensifying screen
C
Film base with emulsion
on both sides
Silver halide in gelatin on base
Bottom intensifying screen
Cassette casing
Intensifying screen on base
FIGURE 7-4 A. A diagram showing the various layers in a double-sided screen-film cassette is shown, with
the dual emulsion film laying inside the screen with two intensifying screens. B. A micrograph of the intensifying screen phosphor, attached to a plastic base, is shown. The scintillating layer consists of small (,25 mm)
crystals of phosphor (i.e., Gd2O2S:Eu), packed with a binder into a screen. This is an unstructured or amorphous
intensifying screen. C. A micrograph of the emulsion layer is shown. The individual grains of silver halide are
seen within the gelatin layer.
Screens
In screen-film radiography, a sheet of film with a light-sensitive emulsion on both
sides is sandwiched between two intensifying screens (Fig. 7-4). A light-tight cassette encapsulates the screens and film. The intensifying screens are composed of
a scintillator, such as Gd2O2S crystals, held together by a binder material (Fig. 7-4
shows a micrograph of a screen). The scintillator in the intensifying screen converts incident x-ray photons to visible light, which then exposes the silver halide
emulsion on the film. Close physical contact between the screen and film is very
important, as gaps promote excessive lateral light propagation, which blur the image.
The screens in a cassette are mounted on compressible foam backing, which presses
the screens against the film to maintain good screen-film contact. The intensifying
screen is composed of high Z compounds that have high absorption efficiency for
x-rays. CaWO4 initially was used; more common today are the rare-earth scintillators
Gd2O2S, LaOBr, and YTaO4. Thicker intensifying screens absorb more x-ray photons
than thinner screens, but the scintillation light that is emitted in a thicker screen diffuses greater distances to reach and expose the film emulsion (Fig. 7-5). This results
in more light spread, which reduces the spatial resolution of the screen. For very
high-resolution screen-film systems (some extremity cassettes and nearly all mammography systems), only one screen is used in conjunction with a single emulsion
film. Intensifying screens are discussed in more detail later in this chapter.
Film
Film is composed of a thin plastic base coated on one or both sides with a layer of
light-sensitive emulsion consisting of silver halide (about 95% AgBr and 15% AgI)
crystals held in water-soluble gelatin (Fig. 7-4 shows a micrograph). The shapes of
the silver halide crystals incorporated in radiographic film are engineered in a tabular
211
FIGURE 7-5 The indirect detection process is illustrated, with a thick screen (A) and a thinner screen (B). X-ray
photons are incident on both screen-film systems, interact in both the upper and lower screens, and the light photons that are created by scintillation then diffuse by scattering through the screen matrix, until some eventually
reach the film emulsion. The thicker screen creates a geometry where the light photons can travel farther laterally,
and this increases the width of the light distribution reaching the emulsion, compared to the thinner screen. Thus,
thicker screens absorb more photons (are more sensitive) but have reduced spatial resolution compared to thinner
screens. Not shown in this diagram is the fact that exponentially more x-ray photons are deposited in the upper
screens as compared to the lower screens, due to the exponential manner in which x-rays are absorbed.
T grain shape (see Fig. 7-4) to increase surface area for improving light capture
efficiency and reducing silver content to save cost. Silver halide grains have a small
amount of AgS that introduces defects in the ionic crystal structures, where the negative charge (from Br and I) builds up on the surfaces and a net positive charge (Ag)
is more central in the crystals. A sensitivity speck, induced by the lattice defects caused
by AgS, is a protrusion of the positive charge that reaches the surface of the crystal.
When a silver halide grain is exposed to visible light, a small number of Ag ions are
reduced (gain electrons) and are converted into metallic Ag (Ag e Ag). If 5
Ag ions are reduced, which depends on the incident light intensity, a stable latent
image center is formed. An exposed film that has not been chemically developed contains a latent image consisting of a pattern of invisible silver halide grains with latent
image centers. There is also a slight intensity rate effect, giving rise to slight differences in film density depending on the exposure rate (called reciprocity law failure).
Film Processing
The gelatin layer is permeable to aqueous solutions, which is necessary for the aqueous developing chemicals to come into contact with the silver crystals. When the
film is developed in an aqueous chemical bath containing a reducing agent, called
the developer, the metallic Ag atoms at the latent image centers act as a catalyst, causing the remaining silver ions in that grain to be reduced. A grain of reduced metallic
silver atoms appears as a black speck on the film.
After the film has passed through the developer, it passes through a bath of an aqueous oxidizing solution called fixer that dissolves the remaining (inactivated) silver halide
from the emulsion layer areas that were not exposed (or were underexposed) to light.
The film is then rinsed with water to remove residual developer and fixer, and is dried.
212
In this way, the visible light striking the film emulsion acts locally to cause film
darkening. The film appears darker to the human eye as a result of an increasing
density of microscopic black silver grains in the emulsion layer. Thus, when the film
is placed on a light box, a gray scale image is produced that is darker where the x-ray
beam incident on the screen-film detector was more intense (where x-ray attenuation in the patient was lower), and the film is lighter where the x-ray beam was less
intense (where the attenuation in the patient was higher). Film radiographs are negative imageshigher exposure to the detector produces darker images.
The details of film processing are outside the scope of this chapter. However, it is
worth noting that the catalytic reaction that is promoted by the sensitivity speck in
the presence of a strong reducing agent results in an approximately 109 gain in optical density on the film with very little added noisethis is a chemical amplification
process that few electronic detector systems can match.
The Characteristic Curve

Although the intensifying screen responds linearly to incident x-ray exposure (e.g.,
twice the x-ray fluence results in twice the visible light output), the response of the
film emulsion to light results in a nonlinear characteristic curve for screen-film systems. An H and D curve (for Hurter and Driffield) shown in Figure 7-6, is a plot
of the log of the relative exposure on the x-axis and the optical density (OD) of the
processed film on the y-axis. This figure illustrates that the optical density (OD) of
film changes nonlinearly with the x-ray exposure level. The OD is defined as the
negative of the base-10 logarithm of the transmittance (T) of the film

OD log10 (T ),
[7-4]
In Eq. 7-4, T is defined as the fraction of visible light passing through the film from
a light source:

I
,
Io
[7-5]
3.5
shoulder
3.0
ion
2.5
ea
rr
eg
2.0
1.5
lin
Optical Density
FIGURE 7-6 A characteristic curve for

screen-film is shown. This is also called the
H and D curve after Hurter and Driffields
early work. The H and D curve is typically
shown as the OD as a function of log-relative
exposure. The curve shows low contrast in
the toe and shoulder regions, and a higher
contrast (higher slope) in the linear region.
If an unexposed film is developed, its OD
will be the base + fog level.
1.0
base + fog
0.5
toe
0.0
1.0
3.0
10.0
Kerma (Gy)
30.0
100.0
213
In Eq. 7-5, I is the intensity of light that passes through the film and Io is the intensity
of the light source at the same point with the film removed. From the above equations,
a film of OD1 transmits 10%, OD2 transmits 1% and OD3 transmits 0.1% of
the intensity of a light source from a fluorescent light box. Since the human eye
responds approximately logarithmically to light intensity, a film OD of 1.0 appears
to be two times brighter than an OD of 2.0, and three times brighter than an OD of
3.0. The inverse of the relationship given in Eq. 7-4 is given by:
10OD T .
[7-6]
Radiography has many clinical applications, and both the screen characteristics
and film characteristics can be changed to optimize the detector for different clinical
demands. Screen-film sensitivity can be adjusted as shown in Figure 7-7A. Systems can
be more sensitive (and have higher speed), which means that less x-ray radiation is
required to produce a given OD. Slower screen-film systems require more radiation
to produce a given level of film darkening, which results in more radiation dose, but
because more x-ray photons are used to make the image, the statistical integrity of the
image is better (lower quantum noise). A change in screen-film speed is seen as a lateral shift of the H and D curve (Fig. 7-7A). Screen-film systems can also be adjusted to
deliver different image contrast, as seen in Figure 7-7B. Contrast is adjusted by altering
the grain size and other parameters in the film emulsion. While high contrast is desirable, the compromise is that dynamic range (called latitude in radiography) is reduced.
High contrast films are not suitable for all applications. For example, a large
difference in x-ray fluence to the detector occurs in chest radiography, because of
the highly attenuating mediastinum and the highly transmissive lung fieldsa lowcontrast, wide-latitude film is required for this setting, otherwise the heart silhouette
would be too white and the lung fields would be too dark on the processed film.
Moreover, very high contrast films require more precise exposure levels to maintain
3.5
3.5
3.0
3.0
Latitude B
2.5
Optical Density
Optical Density
A
2.5
2.0
1.5
base + fog + 1.0
1.0
2.0
1.5
1.0
0.5
0.5
0.0
1.0
0.0
1.0
3.0
10.0
Kerma (Gy)
30.0
100.0
Latitude A
3.0
10.0
Kerma (Gy)
30.0
100.0
FIGURE 7-7 A. Two characteristic curves are shown, one for a more sensitive system (curve A), and one for
a system with lower sensitivity (curve B). The speed of a screen-film system relates to how much exposure is
required to deliver a specific OD, typically OD 1 base fog. Curve A requires less radiation to reach this
OD, and hence it is faster. B. With these two characteristic curves, the speed is approximately the same; however, the slopes of these curves are different. Curve A has a steeper slope, and therefore has higher contrast,
but the exposure latitude is narrowthe exposure levels striking the screen-film system are optimally kept
to within this range. Curve B has less slope, and will produce less radiographic contrast, but it has a wider
exposure latitude. For examinations where a lot of subject contrast is expected (such as the chest radiograph),
wider latitude screen-film systems are required.
214
proper optical density on the film, and failure to achieve such precision may require
more studies to be repeated. For mammography, on the other hand, the compressed
breast is almost uniformly thick and thus the x-ray fluence striking the detector
(under the breast anatomy) is relatively uniform, and thus a very high contrast (but
narrow latitude) film can be used (system A in Fig. 7-7B).
The concepts of changing the characteristics of screen film systems are discussed
above, however in actual practice screen film characteristics are fixed by the manufacturer. The user typically discusses the needs of their imaging facility with the
film vendor, and purchases the screen film system or systems appropriate for their
needs.
7.3 Computed Radiography

Computed radiography (CR) refers to photostimulable phosphor detector (PSP) systems, which are historically housed in a cassette similar to a screen-film cassette.
Traditional scintillators, such as Gd2O2S and cesium iodide (CsI), emit light promptly
(nearly instantaneously) when irradiated by an x-ray beam. When x-rays are absorbed
by photostimulable phosphors, some light is also promptly emitted, but a fraction
of the absorbed x-ray energy is trapped in the PSP screen and can be read out later
using laser light. For this reason, PSP screens are also called storage phosphors. CR was
introduced in the 1970s, saw increasing use in the late 1980s, and was in wide use
at the turn of the century as many departments installed PACS.
Most CR imaging plates are composed of a mixture of BaFBr and other halidebased phosphors, often referred to as barium fluorohalide. A CR plate is a flexible
screen that is enclosed in a light-tight cassette. The CR cassette is exposed to x-rays
during the radiographic examination and is subsequently placed in a CR reader.
Once placed in the CR reader, the following steps take place:
1. The cassette is moved into the reader unit, and the imaging plate is mechanically
removed from the cassette.
2. The imaging plate is translated vertically in the (y) direction by rollers across a
moving stage and is scanned horizontally in the (x) direction by a laser beam of
approximately 700 nm wavelength.
3. Red laser light stimulates the emission of trapped energy in a tiny area (x,y location) of the imaging plate, and blue-green visible light is emitted from the storage
phosphor as energetic electrons drop down to their ground state.
4. The light emitted through photostimulated luminescence is collected by a fiber
optic light guide and strikes a photomultiplier tube (PMT), where it produces an
electronic signal.
5. The electronic signal is digitized and stored as a pixel value. For every spatial
location (x, y) on the imaging plate, a corresponding gray scale value is determined that is proportional to the locally absorbed x-ray energy.
6. The plate is exposed to bright white light to erase any residual trapped energy.
7. The imaging plate is returned to the cassette and is ready for reuse.
The digital image that is generated by the CR reader is stored temporarily on a local
hard disk. Once acquired, the digital radiographic image undergoes image processing by the CR reader. The image is then typically sent to a PACS for interpretation by
a radiologist and long-term archiving.
The imaging plate itself is a completely analog device, but it is read out by analog
and digital electronic techniques, as shown in Figure 7-8.
Photomultiplier tube
Rotating mirror
Light filter
Laser diode
Light guide
215
FIGURE 7-8 The readout mechanics

of a CR system are shown. The imaging
plate is translated through the mechanism by a series of rollers, and a laser
beam scans horizontally across the plate.
The rotating multifaceted mirror causes
the laser beam to scan the imaging plate
in a raster fashion. The light released by
laser stimulation is collected by the light
guide and produces a signal in the PMT.
The red light from the laser is filtered out
before reaching the PMT.
Imaging plate
Rollers
Light Intensity (a.u.)
The light that is released from the imaging plate is a different color than the stimulating
laser light (Fig. 7-9). Indeed, the blue light that is emitted from the screen has shorter
wavelength and higher energy per photon than the stimulating red light. This is energetically feasible only because the energy associated with the emitted blue light was actually
stored in the screen from the absorption of very energetic x-ray photons, and the red
light serves only to release this energy from the storage phosphor. To eliminate detection
by the PMT of the scattered excitation laser light, an optical filter that is positioned in
front of the PMT transmits the blue light and attenuates the red laser light.
Figure 7-10 illustrates how photostimulable phosphors work. A small mass of
screen material is shown being exposed to x-rays. Typical imaging plates are composed of about 85% BaFBr and 15% BaFI, activated with a small quantity of europium. The CR screen is amorphous (lacks structure); the barium fluorohalide crystals
are small and are held together by an inert, optically transparent binder. The nomenclature BaFBr:Eu indicates that the BaFBr phosphor is activated by europium. This
activation procedure, also called doping, creates defects in the BaFBr crystals that
allow electrons to be trapped more efficiently. The defects in the crystalline lattice
caused by the europium dopant give rise to so-called F-centers.
100
90
80
70
60
50
40
30
20
10
0
300
400
500
600
Wavelength (nm)
700
800
FIGURE 7-9 The red laser light (600 to 700 nm) is used to stimulate the emission of the blue light (400
to 450nm); the intensity of the blue light is proportional to the x-ray exposure to the region of the detector
illuminated by the laser light.
216
red laser beam
x-rays
latent image
conduction band
e
F+
energy
Eu+2
F+
DE
Eu+3
A. exposure
valence band
B. transport
Eu+3
DE
Eu+2
C. readout
FIGURE 7-10 A. During exposure, x-rays are absorbed in the storage phosphor and some of the electrons
reach the conduction band, where they may interact with an F-center, causing the reduction of europium (Eu)
ions. B. The electrons are trapped in this high-energy, metastable state, and can remain there for minutes
to months. C. During readout, the phosphor is scanned by a red laser, which provides the trapped electrons
enough energy to be excited into the conduction band. Here, some fraction of electrons will drop down to
the valence band, emitting blue light during the transition. When the F-center releases a trapped electron, the
trivalent europium (Eu3) is converted back to its divalent state. (Eu2).
When x-ray energy is absorbed by the BaFBr phosphor, the absorbed energy
excites electrons associated with the europium atoms, causing divalent europium
atoms (Eu2) to be oxidized and changed to the trivalent state (Eu3). The excited
electrons become mobile, and a fraction of them interact with F-centers. The F-centers
trap these electrons in a higher-energy, metastable state, where they can remain for
minutes to weeks, with only slight fading over time. The latent image that is encoded
on the imaging plate after x-ray exposure, but before readout, exists as billions of
electrons trapped in F-centers. The number of trapped excited electrons per unit area
of the imaging plate is proportional to the intensity of x-rays incident at each location
of the detector during the x-ray exposure.
When the red laser light scans the exposed imaging plate, some energy is absorbed
at the F-center and transferred to the electrons. A proportional fraction of electrons gain
enough energy to reach the conduction band, become mobile, and then drop to the
ground energy state, and thus Eu3 is converted back to Eu2. Emission of the blue emitted light is captured by a light guide and is then channeled to the PMT (Fig. 7-10).
The first readout of the imaging plate does not release all of the trapped electrons
that form the latent image; indeed, a plate can be readout a second time and a third
time with only slight degradation. To erase the latent image so that the imaging
plate can be reused for another exposure without ghosting, the plate is exposed to a
very bright light source, which flushes almost all of the metastable electrons to their
ground state, emptying most of the F-centers.
217
7.4 Charge-Coupled Device and Complementary
Metal-Oxide Semiconductor detectors
Charge-coupled device (CCD) detectors form images from visible light (Fig. 7-11A).
CCD detectors are used in commercial-grade television cameras and in scientific applications such as astronomy. The CCD chip itself is an integrated circuit made of crystalline
silicon, as is the central processing unit of a computer. A CCD chip has an array of discrete detector electronics etched into its surface. Linear arrays are configured with single
or multiple rows in a wide selection, such as 1 2048 detector elements (dexels), or 96
4096 dexels in a rectangular format for line scan detection (e.g, movement ofaslotscan detector with a narrow fan-beam collimation). Area arrays have a 2.5 2.5-cm
dimension with 1,024 1,024 or 2,048 2,048 detector elements on their surface.
Larger chips and larger matrices spanning 6 6 cm are available, but are very expensive,
and ultimately the size is limited by the dimensions of crystalline silicon wafers. Another
limitation is the requirement for a small dexel (e.g., 20 mm dimension and smaller) to
achieve charge transfer efficiency of 99.99% to keep additive electronic noise low. The
silicon surface of a CCD chip is photosensitiveas visible light falls on each dexel, electrons are liberated and build up in the dexel. More electrons are produced in dexels that
receive more intense light. The electrons are confined to each dexel because there are
electronic barriers (voltage) on each side of the dexel during exposure.
Once the CCD chip has been exposed, the electrical charge that resides in each
dexel is read out. The readout process is akin to a bucket brigade (Fig. 7-11B).
Along one column of the CCD chip, the electronic charge is shifted dexel by dexel
FIGURE 7-11 A. A photograph of a high-resolution CCD chip is shown. B. The readout procedure in a CCD
chip is illustrated. After exposure, electrodes in the chip shift the charge packets for each detector element
by switching voltages, allowing the charge packets to move down by one detector element at a time. Charge
from the bottom element of each column spills onto the readout row, which is rapidly read out horizontally.
This process repeats itself until all rows in each column are read out. C. This illustration shows the shift of a
given pattern of exposure down one column in a CCD chip in four (t1t4) successive clock cycles.
218
FIGURE 7-12 A. A CCD-based DR

system is shown. The x-ray is converted
to visible light in the intensifying screen,
which propagates through the lighttight enclosure. A small fraction of the
light photons will strike the mirror and
be redirected to the lens to ultimately
be detected in the CCD chip. B. A CCDbased Time Delay and Integration (TDI)
system is shown. As the x-ray tube / slitslot / detector assembly moves at velocity
V (to the left in this figure), the CCD chip
is read-out at a velocity V (to the right).
By synchronizing the velocity of the read
out with the scanning motion, the signal region under an object in the patient
travels across the entire field of view of
the detector, and its contrast signal is
built up across all of the dexels in the TDI
CCD chip.
CCD chip
Intensifying screen
Lens
Xray
Mirror
Xray tube
whole
assembly
moves
prepatient slot
scanning Xray beam
postpatient slit
CCD detector (endon view)
VS
object in patient
slit
TDI CCD
signal shift
by appropriate control of voltage levels at the boundaries of each dexel. The charge
packet from each dexel in the entire column is shifted simultaneously, in parallel. For
a two-dimensional CCD detector, the charges on each column are shifted onto the bottom row of electronics, that entire row is read out horizontally, then the charge packets
from all columns are shifted down one detector element (Fig. 7-11C), and so on.
CCD detectors are small, whereas the field of view in most medical imaging applications is large. Unfortunately, it is impossible to focus the light emitted from a large
scintillation screen (,43 43 cm) onto the surface of the CCD chip (,4 4 cm)
without losing a large fraction of light photons due to optical lens coupling inefficiency.
The amount of light transferred to the CCD is determined by the directionality of light
emitted by the scintillator, the lens characteristics (f number, the ratio of the focal
length of a lens to its effective diameter), and the demagnification factor, m, required to
focus the image from the screen onto the CCD array. For a typical scintillator material
(so-called lambertian emitter), the amount of light recorded is inversely proportional
to the square of the f number times the square of the demagnification factor. Even with
an excellent lens (e.g., f number 1) and a demagnification of 10, less than 1% of the
light produced by x-ray interactions will reach the CCD array. (Fig. 7-12A).
219
Many x-ray imaging systems are multi-stage, where the signal is converted from
stage to stage. For example, in a CCD system, x-rays are converted to visible light in
the scintillator, and then the light is converted to electrons in the CCD chip. X-ray
photons, visible photons, and electrons are all forms of quanta. The quantum sink refers
to the stage where the number of quanta is the lowest and therefore where the statistical integrity of the signal is the worst. Ideally, in a radiographic image, the quantum
sink should be at the stage where the x-ray photons are absorbed in the converter
(scintillator or solid state detector). This is referred to as an x-ray quantum limited
detector. However, if there is a subsequent stage where the number of quanta are less
than the number of absorbed x-rays, the image statistics and consequently the image
noise will be dominated by this stage, and a secondary quantum sink will occur.
Although an image can be produced, an x-ray detector system with a secondary quantum sink will have image quality that is not commensurate with the x-ray dose used to
make the image. Large field of view CCD-based radiographic detectors have a secondary quantum sink due to the very low coupling efficiency of the lens system. Despite
this, higher doses can be used to create acceptable quality images. These systems are
relatively cost-effective and are sold commercially as entry-level digital radiographic
systems. Higher end area CCD radiography systems typically make use of CsI input
screens with improved x-ray detection efficiency and light capture efficiency, while less
costly systems use Gd2O2S screens. For a typical 43 43 cm FOV, 3,000 3,000pixel CCD camera (4 4 cm area with ,13 mm dexels) and ,10:1 demagnification,
the pixel dimensions in the resulting images are approximately 140 to 150 mm.
Linear CCD arrays optically coupled to an x-ray scintillator by fiberoptic channel plates (a light guide made of individual light conducting fibers), often with a
de-magnification taper of 2:1 to 3:1, are used in slot-scan x-ray systems. These systems operate using a narrow x-ray fan beam with pre-and post-patient collimators,
acquiring an image by scanning the beam over the anatomy for several seconds.
An advantage of this geometry is the excellent scatter rejection achieved, allowing
the elimination of the anti-scatter grid and the associated dose penalty (see section
712in this chapter). Coupling of the scintillator to the linear CCD array requires
less demagnification than with a two dimensional CCD system, reducing the
secondary quantum sink problem. A readout process known as Time-Delay and
Integration (TDI) can deliver high dose efficiency and good signal to noise ratio.
TDI readout is a method that electronically integrates signal information from the
stationary patient anatomy by compensating for the velocity V of the x-ray/detector assembly by reading out the CCD in the opposite direction at a velocity V.
(Fig 7-12B). The transmitted x-ray beam passes through the same object path during
the dwell time of the scan over the area. Disadvantages of slot-scan systems include
the longer exposure time required for the scan, increasing the potential for motion
artifacts, and substantially increased x-ray tube loading. Imaging systems using slotscan acquisition have shown excellent clinical usefulness in x-ray examinations of
the chest and in full-body trauma imaging.
Complementary Metal-Oxide Semiconductor (CMOS) light sensitive arrays are
an alternative to the CCD arrays discussed above. Based upon a crystalline silicon
matrix, these arrays are essentially random access memory chips with built-in
photo-sensitive detectors, storage capacitors, and active readout electronics, operating
at low voltage (3 to 5 volts). Inherent in the CMOS design is the ability to randomly
address any detector element on the chip in a read or read and erase mode, enabling
unique opportunities for built-in automatic exposure control (AEC) capabilities that
are not easily performed with a CCD photo detector. A major issue with CMOS
has been electronic noise from both acquisition (storage) and readout / reset noise
220
sources. Correlated sampling methods can reduce the electronic reset noise common
to multiple measurements. Construction of a large area detector is a hurdle, due to
the maximum dexel size currently achievable (on the order of 50 mm). CMOS detector applications for radiography are currently limited to small FOV (10 15 cm)
applications such as specimen tissue imaging of surgery samples.
7.5 Flat Panel Thin-Film-Transistor Array Detectors

Flat panel Thin-Film-Transistor (TFT) array detectors (Fig. 7-13) make use of technology similar to that used in flat panel displays, and much of this has to do with
the wiring requirements of a huge number of individual display elements. Instead
of producing individual electrical connections to each one of the elements in a flat
panel display, a series of horizontal and vertical electrical lines is used which, when
combined with appropriate readout logic, can address each individual display
element. This signal modulates light transmittance from a backlit liquid crystal display element in the flat panel display. With this approach, only 2000 connections
between the display and the electronics are required for a 1000 1000 display,
instead of 1,000,000 individual connections. For a flat panel display, the wiring
is used to send signals from the computer graphics card to each display element,
whereas in an x-ray detector the wiring is used to measure the signal generated in
each detector element.
Flat-panel TFT arrays are made of amorphous silicon, where lithographic etching
techniques are used to deposit electronic components and connections necessary for
x-ray detector operation. The large area TFT array is divided into individual detector elements (dexels), arranged in a row and column matrix. Electronic components
Detector array
Discrete
detector
elements
Electronics
Light sensitive area
Detector element
FIGURE 7-13 A. flat panel detector systems are pixelated descrete detector systems. The detector array is
comprised of a large number of individual detector elements (dexels). Each dexel has a light sensitive region
and a light-insensitive area where the electronic components are located. B. A photomicrograph of an actual
TFT system is shown. The electronics component can be seen in the upper left corner of each dexel (Image
courtesy John Sabol and Bill Hennessy, GE Healthcare).
221
within each dexel include a TFT, a charge collection electrode, and a storage capacitor.
The TFT is an electronic switch that is comprised of three connections: gate, source,
and drain. Gate and drain lines connect the source and drain of the TFTs along the
row and columns, respectively.
The gate is the transistors on off switch, and is attached to the gate conductor
line along each row of the array. The source is attached to the storage capacitor, and
the drain is attached to the drain conductor line running along each column of the
array. The charge collection electrode captures the charge produced by incident x-ray
energy deposited over the area of the dexel (either by indirect or direct conversion, as
discussed below), and the storage capacitor stores it. During x-ray exposure, the TFT
switch is closed, allowing charge in each dexel to be accumulated and stored. After
the exposure is completed, sequential activation of the TFT array occurs one row at a
time, by sequentially turning on the gate line to every dexel in the row (Fig7-14). This
allows the accumulated charge in each dexel capacitor to flow through the transistor
to the drain line, and subsequently to the connected charge amplifier. The charge
amplifiers are positioned outside of the panel active area. They amplify the charge,
convert it to a proportional voltage, and digitize the voltage level, resulting in a gray
scale value for each dexel in the row. This sequence is repeated row by row, to fully
read out the array. The speed of the detector readout is governed by the intrinsic electronic characteristics of the x-ray converter material and the TFT arrayelectronics.
Indirect Detection TFT Arrays

Indirect x-ray conversion TFT arrays use a scintillator to convert x-rays to light with
optical coupling of the scintillator to the active matrix. The scintillator is layered on
the front surface of the flat panel array. Thus the light emanating from the back of
the scintillator layer strikes the flat panel and, because x-ray interactions are more
likely to occur toward the front of the scintillator layer, the light that is released in
the scintillation layer has to propagate relatively large distances, which can result
in appreciable blurring and a loss of spatial resolution (Fig. 7-15A). To improve
Multiplexer
S1
S2
S3
Charge amplifer
Gate
lines
Drain lines
FIGURE 7-14 This diagram shows circuitry for a TFT flat panel detector system. The readout process is
described in the text.
222

Xray photon
light photons
TFT detector
elements ion pairs
Xray photon
electric field
detected
signal
INDIRECT DETECTION
DIRECT DETECTION
FIGURE 7-15 Indirect and direct detector TFTbased x-ray detectors are shown. A. Photons in the indirect system propagate laterally, compromising resolution. The detected signal shown for the indirect detector shows this
lateral spread in the signal from one x-ray photons interaction. B. For the direct detector system, the ion pairs liberated by x-ray interaction follow the electric field lines (electron holes travel upwards, electrons travel downwards)
and have negligible lateral spread. Here, the detected electronic signal from one x-ray photons interaction is
collected almost entirely in one detector element, and therefore better spatial resolution is achieved.
this situation, most flat panel detector systems for general radiography use a CsI
scintillator instead of Gd2O2S. CsI is grown in columnar crystals, and the columns
act as light pipes to reduce the lateral spread of light. The reduction of the lateral
propagation of light helps to preserve spatial resolution.
Because the electronics requires a certain amount of area on the dexel, the entire
surface area is not photosensitive. This reduces the geometrical efficiency of light collection of each dexel to less than 100%. The fill factor refers to the percent of the area
of each dexel that is photosensitive. The fill factor is typically about 80% for dexel
dimensions of 200 mm 200 mm, and much less (40% to 50%) for smaller dexel
dimensions. Fill factor issues place limitations on how small dexels in a TFT array can
be made, and currently the smallest indirect detector dexel size is about 100mm. Technological advances are now overcoming fill-factor penalties by stacking the electrical
components of each dexel in lithographic layers under the p
hotosensitivelayer.
Direct Detection TFT Arrays

Direct x-ray conversion TFT arrays use a semiconductor material that produces
electron-hole pairs in proportion to the incident x-ray intensity. Absorbed x-ray
energy is directly converted into charge in the detector - there is no intermediate step
involving the production of visible light photons. Amorphous selenium (a-Se) is the
semiconductor most widely used, and is layered between two surface-area electrodes
connected to the bias voltage and a dielectric layer. The dielectric layer prevents overcharging dexels, which could damage the TFT array. Ion pairs are collected under
applied voltage across the solid state converter (1050 V/mum of thickness). This
electric field in the converter almost completely eliminates lateral spreading of the
charges during transit through the semiconductor, resulting in high spatial resolution
(Fig. 7-15B). Even though Se has a relatively low atomic number and consequently
low absorption efficiency, the Se layer can be made thick (0.5 to 1.0 mm) to improve
detection efficiency and still maintain excellent spatial resolution. Fill factor penalties are not as significant with direct conversion TFT systems compared to indirect
conversion TFT arrays. This is because the electrical potential field lines are designed
to bend and thereby direct charge carriers to the collection electrode, avoiding the
insensitive regions of the dexel. In addition to selenium, other materials such as
mercuric iodide (HgI2), lead iodide (PbI2), and cadmium telluride (CdTe) are being
studied for use in direct detection flat panel systems.
223
7.6 Technique Factors in Radiography

The principal x-ray technique factors used for radiography include the tube voltage
(the kV), the tube current (mA), the exposure time, and the x-ray source-to-image
distance, SID. The SID is standardized to 100 cm typically (Fig. 7-16), and 183 cm
for upright chest radiography. In general, lower kV settings will increase the dose
to the patient compared to higher kV settings for the same imaging procedure and
same body part, but the trade-off is that subject contrast is reduced with higher kV.
The kV is usually adjusted according to the examination typelower kVs are used
for bone imaging and when iodine or barium contrast agents are used; however, the
kV is also adjusted to accommodate the thickness of the body part. For example, in
bone imaging applications, 55 kV can be used for wrist imaging since the forearm is
relatively thin, whereas 75 to 90 kV might be used for L-spine radiography, depending on the size of the patients abdomen. The use of the 55-kV beam in abdominal
imaging would result in a prohibitively high radiation dose. Lower kVs emphasize
contrast due to the photoelectric effect in the patient, which is important for higher
atomic number (Z) materials such as bone (Zcalcium 20) and contrast agents containing iodine (Z 53) or barium (Z 56). Conversely, for chest radiography, the soft
tissues of the cardiac silhouette and pulmonary anatomy are of interest, and the ribs
obscure them. In this case, high kV is used (typically 120 kV) in order to decrease
the conspicuity of the ribs by reducing photoelectric interactions.
With the SID and kV adjusted as described above, the overall x-ray fluence is
then adjusted by using the mA and the time (measured in seconds). The product of
the mA and time (s) is called the mAs, and at the same kV and distance, the x-ray
fluence is linearly proportional to the mAsdouble the mAs, and the x-ray fluence
doubles. Modern x-ray generators and tubes designed for radiography can operate
at relatively high mA (such as 500 to 800 mA), and higher mA settings allow the
exposure time to be shorter, which freezes patient motion.
Manual x-ray technique selection was used up until the 1970s in radiography,
whereby the technologist selected the kV and mAs based on experience and in-house
technique charts. Some institutions required technologists to use large calipers to
measure the thickness of the body part to be imaged, to improve the consistency of
the radiographic film optical density. Today, radiography usually makes use of automatic exposure control (AEC), which informally is called phototiming. The mA is set to
a high value such as 500 mA, and the exposure time (and hence the overall mAs) is
determined by the AEC system. In general radiography, an air-ionization photocell
is located behind the patient and grid, but in front of the x-ray detector (Fig.7-16).
100 cm
Xray tube
aluminum filters
colimator
antiscatter grid
photocell
Xray detector
FIGURE 7-16 The standard configuration

for radiography is illustrated. Most tablebased radiographic systems use a SID of
100 cm. The x-ray collimator has a light
bulb and mirror assembly in it, and (when
activated) this casts a light beam onto the
patient that allows the technologist to position the x-ray beam relative to the patients
anatomy. The light beam is congruent with
the x-ray beam. The x-rays that pass through
the patient must pass through the antiscatter
grid and the photocell (part of the AEC system) to then strike the x-ray detector.
224
During exposure, the AEC integrates the exposure signal from the photocell in real
time until a preset exposure level is reached, and then the AEC immediately terminates the exposure. The preset exposure levels are calibrated by the service personnel, and the sensitivity of the radiographic detector is taken into account in the
calibration procedure. For screen-film radiography, the AEC is set to deliver proper
film darkening. For a digital radiographic system, the AEC is adjusted so that the
exposure levels are both in the range of the detector system and produce images with
good statistical integrity based on a predetermined signal-to-noise ratio. This means
that a trade-off between radiation dose to the patient and noise in the digital image
should be reached. For some digital imaging systems, the signal from the detector
itself can be used as the AEC sensor.
7.7 Scintillators and Intensifying Screens

An indirect x-ray detector system uses a scintillator to convert the x-ray fluence incident on the detector into a visible light signal, which is then used to expose the emulsion in screen-film radiography or a photoconductor in digital radiographic systems.
Over the long history of screen-film radiography, a number of scintillators were developed. Most intensifying screens are comprised of fine-grain crystalline scintillating
powders (also called phosphors), formed into a uniformly thick intensifying screen
that is held together by a binder. During production, the phosphor power is mixed
with the binder material at high temperature, and then this molten mixture is layered
onto a metallic sheet, where rollers spread the material into a uniform thickness. The
material hardens as it cools, and the intensifying screen is cut to the desired dimensions. The binder is usually a white powder, and intensifying screens appear white to
the naked eye. Intensifying screens are considered turbid mediathey are not strictly
transparent to optical light (in such case they would be clear in appearance), but
instead the visible light that is produced in the screen from x-ray interactions undergoes many scattering events as it propagates through the intensifying screen. The light
photons have a certain probability of being absorbed as well. Most intensifying screens
are amorphous (lack structure), and are comprised of billions of tiny crystals of the
scintillator randomly embedded in the inert binder layer. Because the index of refraction of the binder is different than the scintillator crystals, the facets of these tiny crystals are sites where light scattering (and absorption) occurs. Each light photon that is
produced by x-ray interaction inside a scintillator crystal propagates in any direction
in the screen, refracting off thousands of small surfaces until it either exits the screen
matrix or is absorbed. Consequently, a burst of perhaps 2,000 light photons produced
at the point of x-ray interaction propagates inside the screen matrix undergoing many
scattering and absorption events, and eventually about 200 photons reach the surface
of the screen where they can interact with the photodetector.
In thicker screens, the packet of photons diffuses a greater thickness before emerging from the phosphor layer, and this results in more spreading of light reaching the
light detector (see Fig. 7-5). The extent of lateral light spread is also governed by the
design of the phosphor screen itselfthe relative coefficients of optical scattering
and optical absorption affect the amount of blurring, and the overall sensitivity of
the screen. Increasing the optical absorption reduces light spread, improving spatial
resolution for the same screen thickness, but less light reaches the surface.
Some crystalline scintillators such as CsI form in long columnar crystals, which
are tiny natural light pipes, and this structured crystal tends to reduce lateral light
spread and preserve spatial resolution for a given screen thickness. CsI is a salt,
225
chemically similar to NaCl, and is hygroscopic (absorbs moisture from the air). It
therefore must be sealed to keep out moisture; it may be kept in a vacuum or encapsulated in a plastic layer. CsI is too fragile and expensive for screen-film radiography
applications.
7.8 Absorption Efficiency and Conversion Efficiency

Because of the dynamics of indirect x-ray detection described above, there are two
factors that are important in x-ray detectors. The first is the absorption efficiency in
the phosphor layer, which is determined by the phosphor composition (its effective
Z and density), and the screen thickness. The absorption efficiency is also dependent
on the x-ray beam energy. The term quantum detection efficiency is used to describe
how well x-ray detectors capture the incident x-ray photon beam, but because this
metric does not consider x-ray fluorescence (where a considerable fraction of the
detected photon energy is re-emitted by the detector), the energy absorption efficiency
is perhaps a better metric, given that most x-ray detectors are energy integrators, not
photon counters. Thicker, denser phosphor layers absorb more of the incident x-ray
beam (Fig. 7-17), and increased absorption efficiency is always desirable. However,
increasing the screen thickness has to be balanced with the concomitant increase in
blurring and loss of spatial resolution that results in thicker screens.
A second factor relates to how efficiently the optical signal is conveyed from the
scintillator to the silicon photodetector, and how it is then amplified and converted
to signal in the image (i.e., gray scale value or digital number). Conversion Efficiency
(CE) includes x-ray to light photon energy transfer in scintillators, the light to charge
fewer incident X-rays

lower dose
higher electronic gain

same screen thickness
X-ray photons
standard dose
for comparison
Increased conversion efficiency
fewer incident X-rays

lower dose
thicker screen
same electronic gain
Increased absorption efficiency
FIGURE 7-17 The center panel is used for comparison, and here a higher exposure level is used as represented by the larger number of x-ray photons incident on the detector system. By increasing the absorption
efficiency (right panel) using a thicker screen by a factor G (G.1), the x-ray beam can be reduced to 1/G
and the same number of x-ray photons will be absorbed in the detector, yielding the same image noise as
before. By increasing the conversion efficiency (left panel) by a factor G, the x-ray beam intensity is turned
down to 1/G to compensate but because the absorption efficiency has not changed in this case, the number
of absorbed photons is decreased, and this increases the image noise.
226
conversion efficiency for photodiodes, the light to OD in film, geometry and lens
efficiency in optically coupled detectors, and the electronic gain in a digital detector
(see Fig. 7-17). To reduce the radiation dose (i.e., use fewer photons to make the
image), but to keep the same gray scale (or OD) in the image, if the conversion efficiency is increased by a factor G (where G . 1) for a system with the same absorption
efficiency, the x-ray fluence must be reduced by a factor of G to deliver the same gray
scale value. This would result in 1/G of the photons being absorbed in the detector,
and hence a G increase in relative quantum noise would result.
If, on the other hand, the absorption efficiency is increased by the factor G, the
x-ray fluence is decreased by the AEC by this same factor. The increase in absorption is balanced by the reduction in x-ray fluence (saving dose), in order to get the
same gray scale. In this case, the same number of x-rays are absorbed by the detector
(i.e., G 1/G 1), and the quantum noise therefore is not changed. So, for a
discussion where the same signal levels in the image are being compared, increasing the
speed of the detector (and lowering the radiation dose) by increasing the conversion
efficiency increases image noise, whereas increasing absorption efficiency can reduce
dose but has no detrimental effect on quantum noise in the image.
7.9 Other Considerations

The choice of a digital detector system requires a careful assessment of the needs of
the facility. CR is often the first digital radiographic system installed in a hospital,
because it can directly replace screen-film cassettes in existing radiography units
and in portable, bedside examinations, where the cost of retakes in both technologist time and money is high. A benefit of CR over TFT-based digital radiographic
technology is that the relatively expensive CR reader is stationary, but because the
CR imaging plates themselves are relatively inexpensive, several dozen plates may
be used in different radiographic rooms simultaneously. The number of rooms
that one CR reader can serve depends on the types of rooms, workload, and proximity. Typically, one multi-cassette CR reader can handle the workload of three
radiographic rooms. The cost of an individual flat panel detector is high, and it
can only be in one place at one time. The obvious use for flat panel detectors or
other solid-state digital radiographic systems is for radiographic suites that have
high patient throughput, such as a dedicated chest room. Because no cassette handling is required and the images are rapidly available for technologist approval, the
throughput in a single room using flat panel detectors can be much higher than
for CR systems, where manual cassette handling is required. Recent introduction
of portable TFT flat panel cassettes using wireless technology is an alternative to
the portable CR detectors that have historically been used in portable radiography,
however, the consequences of the technologist dropping the fragile flat panel cassette can be expensive.
7.10 Radiographic Detectors, Patient Dose, and Exposure
Index
In traditional screen-film radiography, film OD serves as an exposure indicator,

and direct feedback is obtained by simple visual inspection of the processed film
image (Fig. 7-18, left). CR and DR detectors have wide exposure latitude and with
image post-processing, these systems produce consistent image gray scale even with
Fixed Speed
Screen-Film
Digital
Variable Speed
under exposure
high speed
proper exposure
average speed
over exposure
low speed
Exposure distribution
Resultant Images
227
Histogram distribution
FIGURE 7-18 Response of a screen-film detector with fixed radiographic speed and a digital detector
with variable radiographic speed to underexposure, correct exposure, and overexposure are shown. For
screen-film, to the left of each radiograph is a distribution representing the x-ray exposure absorbed by
the screen and converted to OD on the film. Note that the characteristic curve translating exposure to OD
is fixed on the exposure axis. For digital detectors, to the right of each radiograph is a distribution representing the frequency of digital values (a histogram) linear with exposure absorbed by the detector and
converted to a digital value. Note that the look-up table curve translating the digital values to brightness
and contrast on a display monitor is variable and adjusts to the histogram to achieve optimal rendering of
the imagecontrast.
underexposed and overexposed images (Fig. 7-18, right). With the capability of
automatically adjusting the gray scale on digital radiographic images, the checks and
balances provided to the technologist from the immediate feedback of film density
are lost. Underexposed DR images use fewer absorbed x-rays, and can be recognized
by increased image noise, but overexposed images can easily go unnoticed, resulting
in unnecessary overexposure to the patient. Furthermore, underexposed images are
likely to be criticized by radiologists for excessive image noise, whereas overexposed
images will likely be of high quality; this may lead to a phenomenon known as
dose creep, whereby technologists tend to use unnecessarily high exposures. Dose
creep is most likely to occur in examinations such as portable radiography in which
automatic exposure control is not feasible and manual technique factors must be
chosen.
In most digital radiographic systems, image-processing algorithms are used
to align measured histogram values (after exposure) with a predetermined lookup table, to make image gray scale appear similar to screen-film images. The
measured histogram distribution on each radiograph is used to determine the
incident radiation exposure to the detector, and to provide an exposure index
value. Anatomically relevant areas of the radiograph are segmented (parts of the
image corresponding to no patient attenuationhigh values, and collimated
regionslow values, are excluded), a histogram is generated from the relevant
image area, and it is compared to an examination-specific (e.g. chest, forearm,
head, etc.) histogram shape. The gray scale values of the raw image are then
digitally transformed using a look-up-table (LUT) to provide desirable image
contrast in the for presentation image.
228
TABLE 7-1MANUFACTURER, CORRESPONDING SYMBOL FOR EI, AND
ALCULATED EI FOR THREE INCIDENT EXPOSURES TO THE

C
DETECTOR
Manufacturer
SYMBOL
50 Gy
100 Gy
Canon (Brightness = 16, contrast = 10)
REX
50
100
200 Gy
200
Fuji, Konica
400
200
100
Kodak (CR, STD)
EI
1,700
2,000
2,300
IDC (ST = 200)
F#
Philips
EI
200
100
50
Siemens
EI
500
1,000
2,000
The median value of the histogram is used in many systems to determine a

roprietary exposure index (EI) value, which is dependent on each manufacturers
p
algorithm and detector calibration method. This exposure index indicates the amount
of radiation reaching the detector and is not an indicator of dose to the patient.
Unfortunately, widely different methods to calculate the EI value have evolved, as
shown in Table 7-1.
An international standard for an exposure index for digital radiographic systems has been published by the International Electrotechnical Commission (IEC),
IEC 62424-1. This standard describes Exposure Indices and Deviation Indices,
along with a method for placing these values in the DICOM header of each radiographic image. The manufacturers responsibility is to calibrate the imaging detector
according to a detector-specific procedure, to provide methods to segment pertinent
anatomical information in the relevant image region, and to generate an EI from histogram data that is proportional to detector exposure.
This dose index standard for radiography requires staff to establish target exposure
index (EIT) values for each digital radiographic system and for each type of exam; for
instance, the EIT for a skull radiographic will, in general, differ from that of a chest
radiograph. Feedback to the user on whether an appropriate exposure has been
achieved is given by the deviation index (DI), calculated as

DI 10 log10 (EI / EI T )
[7-6]
The DI provides feedback to the operator with a value that is equal to 0 (zero) when
the intended exposure to the detector is achieved (i.e., EI EIT), a positive number
when an overexposure has occurred, and a negative number when an underexposure
has occurred. A DI of 1 indicates an overexposure of about 26%; a value of 1 indicates an underexposure of 20% less than desired. The acceptable range of DI values is
approximately from 1 to 1. When the DI is in the desired range, the radiographic
system is considered to be working well and is able to deliver the EIT values set up by
the institution. Tracking DI values with respect to equipment and technologist can be
useful in maintaining high image quality for radiography at an institution.
7.11 Dual-Energy Radiography

As mentioned in Chapter 4, anatomical noise can contribute to a loss of conspicuity in radiological imaging. Dual-energy radiographic techniques have been developed, which can in some instances reduce anatomical noise and thereby produce an
229
image that has better information content. Dual-energy chest radiography is a good
example of this and will be used to illustrate the concept.
Figure 7-19A shows the attenuation characteristics of iodine, bone, and soft
tissue as a function of x-ray photon energy. The curves have different shapes, and
thus the energy dependencies of bone, iodine, and soft tissue attenuation are different. The reason for these differences is that higher atomic number materials
(i.e., bone, with effective Z < 13, or iodine, with Z 53) have higher photoelectric absorption levels than lower atomic materials such as soft tissue (effective Z <7.6). The photoelectric interaction component of the linear attenuation
coefficient has an energy dependency of E3, as opposed to very little energy
dependency for the Compton scattering component in the x-ray energy range
used for diagnostic imaging. Because of the different energy dependencies in tissue types, the acquisition of two radiographic images at two different effective
energy levels (i.e., different kVs), enables either the bone component or the soft
tissue component in the image to be removed by image post-processing. The
general algorithm for computing dual-energy subtracted images on each pixel
(x,y) is
DE ( x , y ) a b Ln {(I HI ( x , y )} R Ln {I LO ( x , y }
[7-7]
1000
100
10
iodine
bone
soft tissue
0.1
0
20
40
60
80
100
120
Energy (keV)
FIGURE 7-19 Dual-energy chest radiography is illustrated. A. The linear attenuation coefficients for soft tissue, bone, and iodine are shown. B. The single-energy (120 kV) radiograph is shown. C. The bone-subtracted,
soft tissueonly image is illustrated. This image shows the lung parenchyma and mediastinum, which are the
principal organs of interest on most chest radiographs. D. The soft tissuesubtracted, bone-only image is illustrated. Note that the earrings, which are made of metal and are more similar to bone than soft tissue in terms
of composition, show up on the bone-only image.
230
where DE(x,y) is the gray scale value in the dual-energy image, IHI(x,y) is the
c orresponding gray value in the acquired high-energy image, and ILO(x,y) is the corresponding gray scale value in the acquired low-energy image. The value of R is
selected to isolate bone, iodine, or soft tissue in the subtraction, and a and b scale
the output image for brightness and contrast for optimal display. Equation 7-7 represents logarithmic weighted subtraction.
Figure 7-19B shows a standard single-energy chest radiograph for comparison,
and Figure 7-19C and 7-19D illustrate the soft tissue and bone-only images, respectively. Notice on the tissue-only image (Fig. 7-19C), the ribs are eliminated and the
soft tissue parenchyma of the lungs and mediastinum are less obscured by the ribs.
On the bone-only image, bones are of course better depicted. Interestingly, one can
also see the metallic earrings on this patient, whereas they are not visible on the soft
tissueonly image. The metal earrings have attenuation properties more like bone
than soft tissue.
7.12 Scattered Radiation in Projection Radiographic
Imaging
The basic principle of projection x-ray imaging is that x-rays travel in straight lines.
However, when x-ray scattering events occur in the patient, the resulting scattered
x-rays are not aligned with the trajectory of the original primary x-ray, and thus the
straight-line assumption is violated (Fig. 7-20). Scattered radiation that does not
strike the detector has no effect on the image; however, scattered radiation emanating from the patient is of concern for surrounding personnel due to the associated radiation dose. If scattered radiation is detected by the image receptor, it does
have an effect on the image and can be a significant cause of image degradation.
Scattered radiation generates image gray scale where it does not belong, and this
can significantly reduce contrast in screen-film radiography. Because contrast can
be increased in digital images by window/leveling or other adjustment schemes, on
FIGURE 7-20 This illustration demonstrates that when x-rays are

scattered in the patient and reach the detector, they stimulate gray
scale production, but since they are displaced on the image, they carry
little or no information about the patients anatomy.
231
digital radiographic images, scatter acts chiefly as a source of noise, degrading the
signal-to-noise ratio.
The amount of scatter detected in an image is characterized by the scatter-to-primary
ratio (SPR) or the scatter fraction. The SPR is defined as the amount of energy deposited
in a specific location in the detector by scattered photons, divided by the amount of
energy deposited by primary (non-scattered) photons in that same location. Thus,

SPR
S

P
[7-8]
For an SPR of 1, half of the energy deposited on the detector (at that location)
is from scatterthat is, 50% of the information in the image is largely useless. The
scatter fraction (F) is also used to characterize the amount of scatter, defined as

S

P S
The relationship between the SPR and scatter fraction is given by
F
[7-9]
SPR

[7-10]
SPR 1
The vast majority of x-ray detectors integrate the x-ray energy and do not count
photons (which is typical in nuclear medicine). Thus, the terms S and P in this discussion refer to the energies absorbed in the detector from the scattered and primary
photons, respectively.
If uncorrected, the amount of scatter on an image can be high (Fig. 7-21). The SPR
increases typically as the volume of tissue that is irradiated by the x-ray beam increases.
Figure 7-21 illustrates the SPR as a function of the side of a square field of view, for
different patient thicknesses. The SPR increases as the field size increases and as the
thickness of the patient increases. For a typical 30 3 30 abdominal radiograph in a
25-cm-thick patient, the SPR is about 4.5so 82% (the scatter fraction) of the information in the image is essentially useless, if scatter correction methods are not used.

The Anti-scatter Grid

The anti-scatter grid, or sometimes just called a scatter grid, is the most widely used
technology for reducing scatter in radiography, fluoroscopy, and mammography. The
FIGURE 7-21 The SPR is shown as
a function of the side dimension of a
square field of view, for three different patient thicknesses. For example,
the 20-cm point on the x-axis refers
to a 20 3 20-cm field. The SPR
increases with increasing field size
and with increasing patient thickness.
Thus, scatter is much more of a problem in the abdomen as compared to
extremity radiography. Scatter can be
reduced by aggressive use of collimation, which reduces the field of view
of the x-ray beam.
7
30 cm
Scatter to Primary Ratio
6
5
4
20 cm
3
2
10 cm
1
0
0
10
15
20
Field of View (cm)
25
30
232
FIGURE 7-22 The antiscatter grid is located between the patient,

who is the principal source of scatter, and the detector. Grids are geometric devices, and the interspace regions in the grid are aligned with
the x-ray focal spot, which is the location where all primary x-ray photons originate from. Scattered photons are more obliquely oriented,
and as a result have a higher probability of striking the attenuating
septa in the grid. Hence, the grid allows most of the primary radiation
to reach the detector, but prevents most of the scattered radiation from
reaching it.
grid is placed between the detector and the patient (Fig. 7-22). Ideally, it would allow
all primary radiation incident upon it to pass, while absorbing all of the scattered
radiation. The scatter grid has a simple geometric design, in which open interspace
regions and alternating x-ray absorbing septa are aligned with the x-ray tube focal
spot. This alignment allows x-ray photons emanating from the focal spot (primary
radiation) to have a high probability of transmission through the grid (thereby reaching the detector), while more obliquely angled photons (scattered x-rays emanating
in the patient) have a higher probability of striking the highly absorbing grid septa
(Fig. 7-23). The alignment of the grid with the focal spot is crucial to its efficient
FIGURE 7-23 The basic dimensions of a 10:1 antiscatter grid are

shown. The grid is manufactured
from alternating layers of interspace material and septa material. This illustration shows parallel
grid septa; however, in an actual
grid, the septa and interspace
are focused (and thus would be
slightly angled).
Grid Ratio =
H
W
0.12 mm
0.045 mm
1.20 mm
grid septa
interspace material
233
operation, and errors in this alignment can cause reduced grid performance or artifacts in the image.
There are several parameters that characterize the anti-scatter grid. In the clinical
imaging environment (practical radiographic imaging), the following grid parameters
are important:
Grid Ratio
The grid ratio (see Fig. 7-23) is the ratio of the height of the interspace material to
its widththe septa dimensions do not affect the grid ratio metric. Grid ratios in
general diagnostic radiology are generally 6, 8, 10, 12, or 14, with the central three
values used most commonly. Grid ratios are lower (,5) in mammography. The grid
septa in a grid are typically manufactured from lead (z 82, r 11.3 g/cm3). The
grid ratio is the most fundamental descriptor of the grids construction.
Interspace Material
Ideally, air would be the interspace material; however, the lead septa are very malleable and require support for structural integrity. Therefore, a solid material is
placed in the interspace in the typical linear grid used in general radiography to
keep the septa aligned. Low-cost grids in diagnostic radiology can have aluminum
(z 13, r 2.7 g/cm3) as the interspace material, but aluminum can absorb an
appreciable number of the primary photons (especially at lower x-ray energies),
and so, carbon fiber interspaced grids are more common in state-of-the-art imaging systems. Carbon fiber (z 6, r 1.8 g/cm3) has high primary transmission
due to the low atomic number of carbon fiber and its lower density, and thus is a
desirable interspace material.
Grid Frequency
The grid frequency is the number of grid septa per centimeter. Looking at the
grid depicted in Figure 7-23, the septa are 0.045 mm wide and the interspace is
0.120mm wide, resulting in a line pattern with 0.165 mm spacing. The corresponding frequency is 1/0.165mm 6 lines/mm or 60 lines/cm. Grids with 70 or 80 lines/
cm are also available, at greater cost. For imaging systems with discrete detector
elements, a stationary high-frequency grid can be used instead of grid motion (no
Bucky mechanism). For example, a 2,048 3 1,680 chest radiographic system has
approximately 200 mm detector elements, and thus a grid with 45-mm-wide grid
septa (see Fig.7-23) should be largely invisible because the grid bars are substantially
smaller than the spatial resolution of the detector.
Grid Type
The grid pictured in Figure 7-23 is a linear grid, which is fabricated as a series of
alternating septa and interspace layers. Grids with crossed septa are also available but
seldom are used in general radiography applications. Crossed grids are widely used
in mammography (Chapter 8).
Focal Length
The interspace regions in the antiscatter grid should be aligned with the x-ray source,
and this requires that the grid be focused (see Fig. 7-22). The focal length of a grid is
typically 100 cm for most radiographic suites and is 183 cm for most upright chest
imaging units. If the x-ray tube is accidentally located at a different distance from the
grid, then grid cutoff will occur. The focal distance of the grid is more forgiving for
lower grid ratio grids, and therefore high grid ratio grids will have more grid cutoff
234
if the source to detector distance is not exactly at the focal length. For systems where
the source-to-detector distance can vary appreciably during the clinical examination
(fluoroscopy), the use of lower grid ratio grids allows greater flexibility and will suffer less from off-focal grid cutoff, but will be slightly less effective in reducing the
amount of detected scattered radiation.
Moving Grids
Grids are located between the patient and the detector, and for high-resolution
detector systems such as screen-film receptors, the grid bars will be seen on the
image if the grid is stationary. Stationary grids were common in upright screenfilm chest radiography systems, and the success of this approach suggests that
the radiologist is quite adroit at looking through the very regularly spaced grid
lines on the image. ABucky grid is a grid that moves with a reciprocating motion
during the x-ray exposure, causing the grid bars to be blurred by this motion and
not visible in the image. The motion is perpendicular to the long axis of the linear
septa in the grid.
Bucky Factor
The Bucky factor, not to be confused with the moving Bucky grid, describes the relative
increase in x-ray intensity or equivalently, mAs, needed when a grid is used, compared
to when a grid is not used. The Bucky factor essentially describes the radiation dose
penalty of using the gridand typical values of the Bucky factor for abdominal radiography range from 3 to 8. The Bucky factor is relevant in screen-film radiography,
but less so with digital imaging systems. In screen-film radiography, the use of the grid
slightly reduces the amount of detected primary radiation and substantially reduces
the amount of scattered radiation detected, and both of these effects reduce the OD of
the resulting film. Thus, the x-ray technique has to be increased by the Bucky factor
to replace this lost radiation, in order to deliver films of the same OD. Digital systems,
however, have much wider dynamic ranges than screen-film systems in general, and
therefore the x-ray technique does not have to be increased to replace the lost x-ray
exposure to the detector, which was mostly scattered radiation anyway. In practice,
since digital imaging systems have replaced our century-long experience with screenfilm image receptors, the exposures used in digital systems largely follow the trends of
what was used in the screen-film era. Nevertheless, these considerations suggest that
DR systems may be used to provide lower dose examinations than in the screen-film
era.
It should be noted that for thin anatomical structures, there is very little scattered
radiation, and the use of a grid is not necessary. For example, for hand or forearm
radiography, scatter levels are low and the technologist will usually place the limb to
be imaged directly on the detector, with no grid (if it is removable).
There are other parameters that characterize the performance of the anti-scatter
grid, which are used primarily in the scientific evaluation of grid performance. These
parameters are less important to x-ray technologists and radiologists, but are of interest to those engaged in designing grids or optimizing their performance. These metrics are discussed below.
Primary Transmission Factor

Tp is the fraction of primary photons that are transmitted through the grid, and ideally
it would be 1.0. In Figure 7-23, the grid bars cover 27% of the field (0.045/[0.045
0.120]) and primary x-rays striking the top of the septa (parallel to their long axis)
will be mostly attenuated. For that grid, then, the primary transmission would at
235
most be 73%, and this does not consider the attenuation of the interspace material.
Values of Tp typically run from 0.50 to 0.75 with modern grids, and this value is
kV-dependent due to the penetration of the interspace material.
Scatter Transmission Factor

Ts is the fraction of scattered radiation that penetrates the grid. Ideally, it would be
0. The value of Ts can range substantially depending on the amount of scatter in the
field, the x-ray energy, and the grid design. Typical values of Ts run approximately
from 0.05 to 0.20 in general diagnostic radiography.
Selectivity
Selectivity (S) is simply defined as
TP
[7-11]
Contrast Degradation Factor

The contrast degradation factor (CDF) refers to the reduction in contrast due to scattered radiation. It can be shown that contrast is reduced by

CDF
1

1 + SPR
[7-12]
However, because digital images can be adjusted to enhance contrast, CDF is of less
importance in digital radiography than in film-screen radiography.
Other Methods for Scatter Reduction

While the anti-scatter grid is by far the most ubiquitous tool used for scatter reduction, air gaps and scan-slot techniques have been studied for years (Fig. 7-24). The
principle of the air gap is that by moving the patient away from the detector, less of
the scatter that is emitted from the patient will strike the detector. The concept is
FIGURE 7-24 The air gap technique has been used to reduce scattered radiation, and this figure shows the
air gap geometry. The scattered radiation that might strike the detector in the typical geometry (A) has a better probability of not striking the detector as the distance between the patient and the detector is increased.
(B) While often discussed, the air gap technique suffers from field coverage and magnification issues, and is
used only rarely in diagnostic radiography.
236
similar to the inverse square law (radiation intensity decreases as the square of the
distance from the source), except that for an extended source of radiation (scatter
from the patient does not originate from a point, but instead a large volume), the
scattered x-ray intensity with air gap distances is not proportional to 1/r2, but rather
approaches 1/r. Increasing the air gap distance also increases the magnification of
the patients anatomy (see Fig. 7-2). Thus, practical factors limit the utility of the air
gap method for scatter reductionas magnification of the patient anatomy increases,
the coverage of a given detector dimension is reduced, and there is a loss in spatial
resolution due to the increased blurring of the finite focal spot with magnification
(see Fig. 7-3).
The scan-slot system for scatter reduction (Fig. 7-25) is one of the most effective
ways of reducing the detection of scattered x-rays, and the images that result from
scan-slot systems are often noticeably better in appearance due to their excellent
contrast and signal-to-noise ratio. Scan-slot radiography can be regarded as the gold
standard in scatter reduction methods. The method works because a very small field
is being imaged at any point in time (see Fig. 7-21), most scatter is prevented from
reaching the image receptor, and the image is produced by scanning the small slot
across the entire field of view. In addition to excellent scatter rejection, slot-scan radiography also does not require a grid and therefore has the potential to be more dose
efficient. The dose efficiency of scan-slot systems is related to the geometric precision achieved between the alignment of the prepatient slit and the postpatient slot.
If the postpatient slot is collimated too tightly relative to the prepatient slit, primary
radiation that has passed through the patient will be attenuated, and this will reduce
dose efficiency. If the postpatient slot is too wide relative to the prepatient slit, then
more scatter will be detected.
Despite the excellent scatter reduction from these systems, there are limitations. The scanning approach requires significantly longer acquisition times, and the
potential for patient motion during the acquisition increases. The narrow aperture
for scanning requires a high x-ray tube current and the longer scan time requires
significant heat loading of the x-ray tube anode. Alignment of mechanical systems
is always a source of concern, and complex mechanical systems invariably require
more attention from service personnel.
FIGURE 7-25 The scanning

slit (or slot-scan) method for
scatter reduction is illustrated.
The prepatient slit is aligned with
the postpatient slot, and these
two apertures scan across the
field of view together. The challenge is to keep the slot aligned
with the slit, and this can be done
electronically or mechanically.
This geometry was deployed for
an early digital mammography
system, but is also commercially
available for radiography applications. Excellent scatter reduction can be achieved using this
method.
Slit and slot

translate in unison
Slot attenuates forwarddirected scatter from

narrow beam
Pre-patient slit
Patient
Post-patient slot
Detector
237
SUGGESTED READING
Alvarez RE, Seibert JA, Thompson SK. Comparison of dual energy detector system performance, Med Phys
2004;31:556565.
Amis ES, Butler PF, Applegate KE, et al, American College of Radiology White paper on radiation dose in
medicine. JACR 2007;4:272284.
Chan H-P, Doi K. Investigation of the performance of antiscatter grids: Monte Carlo simulation studies,
Phys Med Biol 1982;27:785803.
Chan H-P, Higashida Y, Doi K. Performance of antiscatter grids in diagnostic radiology: Experimental
measurements and Monte Carlo simulations studies, Med Phys 1985;12:449454.
International Electrotechnical Commission, Report IEC 62424-1 (2008).
Shepard SJ, Wang J, Flynn M, et al. An exposure indicator for digital radiography: AAPM Task Group 116
(Executive Summary), Med Phys 2009;36:29892914.
Chapter
Mammography
Mammography is a radiographic examination that is designed for detecting breast
pathology, particularly breast cancer. Breast cancer screening with mammography
assists in detecting cancers at an earlier, more treatable stage, and is an important
clinical procedure because approximately one in eight women will develop breast
cancer over their lifetimes. Technologic advances over the last several decades have
greatly improved the diagnostic sensitivity of mammography. Early x-ray mammography was performed with direct exposure film (intensifying screens were not
used), required high radiation doses, and produced images of low contrast and
poor diagnostic quality. Mammography using the xeroradiographic process was
very popular in the 1970s and early 1980s, spurred by high spatial resolution and
edge-enhanced images; however, its relatively poor sensitivity for breast masses and
higher radiation dose compared to screen-film mammography led to its demise in
the late 1980s. Continuing refinements in screen-film technology and digital mammography, which entered the clinical arena in the early 2000s, further improved
mammography(Fig.8-1).
The American College of Radiology (ACR) mammography accreditation program
changed the practice of mammography in the mid-1980s, with recommendations for
minimum standards of practice and quality control (QC) that spurred improvements
in technology and ensured quality of service. The federal Mammography Quality
Standards Act (MQSA) was enacted in 1992. The law, and associated federal regulations (Title 21 of the Code of Federal Regulations, Part 900) issued by the US Food
and Drug Administration (FDA), made many of the standards of the accreditation
program mandatory. For digital mammography systems, many of the regulatory
requirements entail following the manufacturers recommended QC procedures.
Breast cancer screening programs depend on x-ray mammography because it is a
low-cost, lowradiation dose procedure that has the sensitivity to detect early- stage
breast cancer. Mammographic features characteristic of breast cancer are masses, particularly ones with irregular or spiculated margins; clusters of microcalcifications;
and architectural distortions of breast structures. In screening mammography as practiced in the United States, two x-ray images of each breast, in the mediolateral oblique
and craniocaudal views, are acquired. Whereas screening mammography attempts to
identify breast cancer in the asymptomatic population, diagnostic mammography procedures are performed to assess palpable lesions or evaluate suspicious findings identified by screening mammography. The diagnostic mammographic examination may
include additional x-ray projections, magnification views, spot compression views,
ultrasound, magnetic resonance imaging (MRI), or mammoscintigraphy.
Ultrasound (Chapter 14) is often used to differentiate cysts (typically benign)
from solid masses (often cancerous) and is also used when possible for biopsy needle
guidance. MRI (Chapters 12 and 13) has excellent tissue contrast sensitivity and
with contrast enhancement can differentiate benign from malignant tumors; it is
used for diagnosis, staging, biopsy guidance, and, in some cases, screening. The
clinical utility of mammoscintigraphy utilizing Tc-99m sestamibi is in the evaluation
238
Chapter 8 Mammography
Screen-film (2006)
Digital (2009)
239
Digital tomosynthesis (2010)
FIGURE 8-1 Improvements in mammography. Images of a medial-lateral oblique view of the breast are
shown. On the left is a state-of-the-art screen-film image (2006); in the middle is a corresponding digital mammogram (2008) demonstrating excellent exposure latitude and the skin line not seen in the film image; and
on the right is one of approximately 50 digital tomosynthesis images (2010), which reduce superposition of
anatomy in the projection image at about the same dose as a screen-film mammogram. Today, digital imaging
equipment is used in the majority of mammography clinics.
of suspected breast cancer in patients for whom mammography is nondiagnostic,

equivocal, or difficult to interpret (e.g., the presence of scar tissue, mammographically dense breast tissue, implants, or severe dysplastic disease). It is also used to
assist in identifying multicentric and multifocal carcinomas in patients with tissue
diagnosis of breast cancer.
The morphological differences between normal and cancerous tissues in the
breast and the presence of microcalcifications require the use of x-ray equipment
designed specifically to optimize breast cancer detection. As shown in Figure 8-2A,
the attenuation differences between normal and cancerous tissue are extremely
small. Subject contrast, shown in Figure 8-2B, is highest at low x-ray energies (10
to 15 keV) and reduced at higher energies (e.g., greater than 30 keV). Low x-ray
A
Attenuation of Breast Tissues
B
12
Infiltrating ductal carcinoma
0.3
% subject contrast
Linear attenuation
coefficient (cm1)
1.0
0.5
Glandular
Fat
0.2
0.1
Mammographic Contrast:
Ductal Carcinoma
20
30
50
70
Energy (keV)
100
10
8
6
4
2
0
15
20
25
30
Energy (keV)
35
40
FIGURE 8-2 A. Attenuation of breast tissues as a function of energy, showing fat, glandular, and ductal carcinoma linear attenuation coefficients. Comparison of the three tissues shows a very small difference between the
glandular and the cancerous tissues. B. Calculated percentage contrast of the ductal carcinoma relative to the
glandular tissue declines rapidly with energy; contrast is optimized by the use of a low-energy, nearly monoenergetic x-ray spectrum (Adapted from Yaffe MJ. Digital mammography. In: Haus AG, Yaffe MJ, eds. Syllabus: a
categorical course in physics: technical aspects of breast imaging. Oak Brook, IL: RSNA, 1994:275286.)
240

X-ray tube
Tube port
Filter
Collimator
Compression
paddle
Grid
X-ray detector
Phototimer
detector
FIGURE 8-3 A dedicated mammography system has many unique attributes. Major components of a typical
system, excluding the generator and user console, are shown.
e nergies provide the best differential attenuation between the tissues; however, the
high absorption results in higher radiation doses and long exposure times. Detection
of small calcifications in the breast is also important because microcalcifications are
in some cases early markers of breast cancer. Thus, mammography requires x-ray
detectors with high spatial resolution that function best at higher doses. Enhancing
contrast sensitivity, reducing dose, and providing the spatial resolution necessary to
depict microcalcifications impose extreme requirements on mammographic equipment and detectors. Therefore, dedicated x-ray equipment, specialized x-ray tubes,
breast compression devices, antiscatter grids, x-ray detectors (screen-film or digital),
and phototimer detector systems are essential for mammography (Fig. 8-3). Strict
QC procedures and cooperation among the technologist, radiologist, and medical
physicist are necessary to ensure that high-quality mammograms are achieved at the
lowest possible radiation doses. With these issues in mind, the technical and physical considerations of mammographic x-ray imaging are discussed below. Many of
the basic concepts regarding image quality, x-ray production, and radiography are
presented in Chapters 4, 6, and 7, respectively. The applications of these concepts
to mammography, together with many new topics specific to mammography, are
presented below.
8.1
x-ray Tube and Beam Filtration

A dedicated mammography x-ray tube has more similarities than differences when
compared to a conventional x-ray tube. Therefore, this section highlights the
differences between mammography and general radiography x-ray tubes.
Cathode and Filament Circuit

The mammography x-ray tube is configured with dual filaments in the focusing cup
to produce 0.3- and 0.1-mm focal spot sizes, with the latter used for magnification
studies to reduce geometric blurring, as discussed later in this chapter. An important
241
distinction between mammography and conventional x-ray tube operation is the

low operating voltage, below 40 kV, which requires feedback circuits in the x-ray
generator to adjust the filament current as a function of kV to deliver the desired
tube current because of the nonlinear relationship between filament current and
tube current as described in Chapter 6. In addition, the filament current is restricted
to limit the tube current, typically to 100 mA for the large (0.3 mm) focal spot and
25mA for the small (0.1 mm) focal spot so as to not overheat the Mo or Rh targets
due to the small interaction areas. Higher filament currents and thus tube currents,
up to and beyond 200 mA for the large focal spot and 50 mA for the small focal spot,
are possible with tungsten anodes chiefly due to a higher melting point compared to
Mo and Rh anodes.
Anode
Molybdenum is the most common anode target material used in mammography
x-ray tubes, but Rh and increasingly tungsten (W) are also used as targets. Characteristic x-ray production is the major reason for choosing Mo (K-shell x-ray energies of 17.5 and 19.6 keV) and Rh (20.2 and 22.7 keV) targets, as the numbers of
x-rays in the optimal energy range for breast imaging are significantly increased
by characteristic x-ray emission. With digital detectors, W is becoming the target
of choice. Increased x-ray production efficiency, due to its higher atomic number,
and improved heat loading, due to its higher melting point, are major factors in
favor of W. Digital detectors have extended exposure latitude, and because postacquisition image processing can enhance contrast, characteristic radiation from
Mo or Rh is not as important in digital mammography as it is with screen-film
detectors.
Mammography x-ray tubes have rotating anodes, with anode angles ranging from
16 to 0 degrees, depending on the manufacturer. The tubes are typically positioned
at a source-to-image receptor distance (SID) of about 65 cm. In order to achieve
adequate field coverage on the anterior side of the field, the x-ray tube must be physically tilted so that the effective anode angle (the actual anode angle plus the physical
tube tilt) is at least 22 degrees for coverage of the 24 30-cm field area. A tube with
a 0-degree anode angle requires a tube tilt of about 24 degrees to achieve an effective
anode angle of 24 degrees (Fig. 8-4A). A 16-degree anode angle requires a tube tilt of
6 degrees for an effective angle of 22 degrees (Fig. 8-4B).
The intensity of the x-rays emitted from the focal spot varies within the beam,
with the greatest intensity on the cathode side of the projected field and the lowest
intensity on the anode side, a consequence of the heel effect (Chapter 6). Positioning the cathode over the chest wall of the patient and the anode over the anterior
portion (nipple) achieves better uniformity of the transmitted x-rays through the
breast (Fig.8-5). Orientation of the tube in this way also decreases the equipment
bulk near the patients head. The anode is kept at ground potential (0 voltage),
and the cathode is set to the highest negative voltage to reduce off-focal radiation
(Chapter 6).
Focal Spot Considerations

Focal spot nominal sizes of 0.3 to 0.4 mm for contact mammography (breast compressed against the grid and image receptor) and 0.10 to 0.15 mm for magnification
imaging (breast compressed against a magnification stand, which supports the breast
at a distance from the image receptor to provide geometric image magnification)
242

A
B
16 Anode angle
0 Anode angle
Cathode
6 tube tilt
24 tube tilt
Approximately same field coverage

FIGURE 8-4 Design of a dedicated mammography system includes unique geometry and x-ray tube design.
A half-field geometry is implemented with the x-ray beam central axis perpendicular to the plane of the
image receptor at the center edge of the field at the chest wall, and centered horizontally. To provide full-area
x-ray beam coverage over the large FOV (24 30 cm) at the 60- to 70-cm SID, a tube tilt of about 20 to 24
degrees is necessary. A. An anode angle of 0 degrees requires a tube tilt of 24 degrees. B. An anode angle of
16 degrees requires a tube tilt of about 6 degrees.
Cathode
Chest
wall
Relative photon intensity
FIGURE 8-5 Orientation of the cathode-anode

direction of the x-ray tube is from the chest wall side
of the breast (over the cathode) to the anterior side of
the breast (over the anode). Variation of x-ray beam
intensity is caused by the heel effect, a result of anode
self-filtration of x-rays directed toward the anode side
of the field, with a significant reduction in x-ray intensity. The thickest part of the breast (at the chest wall)
is positioned below the cathode, which helps equalize
x-ray intensities reaching the image receptor that are
transmitted through the breast in the cathode-anode
direction.
Anode
Anterior
100 %
50 %
0%
Cathode
Anode
243
reduce geometric blurring so that microcalcifications can be resolved. A consequence

of using small focal spots is reduced maximal tube current (e.g., for a Mo target,
100 mA for the large focal spot and 25 mA for the small focal spot) and correspondingly longer exposure times.
In order to avoid exposure of the patients torsos and to maximize the amount of
breast tissue near the chest wall that is imaged, all dedicated mammography systems
utilize a half-field x-ray beam geometry, which is achieved by fixed collimation at
the x-ray tube head. As a result, the central axis of the x-ray beam is directed at the
chest wall edge of the receptor and perpendicular to the plane of the image receptor.
Furthermore, by convention, the nominal focal spot size is measured at the reference
axis, which bisects the x-ray field along the chest wallanterior direction of the
x-ray field (Fig. 8-6A).
The sum of the target angle and tube tilt is equal to , and represents the reference angle. At the reference axis, the focal spot length, aref, is foreshortened compared
to the focal spot length at the chest wall, achest wall, given by the relationship
tan(q f )
aref = a chest wall 1
tan q
[8-1]
Nominal focal spot size and tolerance limits for the width and length are listed in
Table 8-1. As a consequence of the effective focal spot size variation in the field (the
line focus principle), sharper image detail is rendered on the anode (anterior) side of
the field toward the nipple, which is more evident with magnification examinations,
as described in Section 8.3.
Tube tilt = 24
65 cm SID
= 12
Actual focal area
Reference
Axis
length
width
= 24
Central
Axis
Projected
focal area
29 cm
FIGURE 8-6 A. The projected focal spot size varies along the cathode-anode axis, and the nominal focal
spot size is specified at a reference axis at an angle f from a line perpendicular to the cathode-anode axis
(dashed line, f 12 degrees). The actual focal area is the electron distribution area on the anode (width
length), and the projected length is foreshortened according to the line focus principle. B. A resolution bar
pattern with object magnification measures overall system resolution including that of the focal spot. Bars of
12 line pairs per mm are resolved for this particular system (magnification 1.5, 0.1-mm focal spot).
244
TABLE 8-1NOMINAL FOCAL SPOT SIZE AND MEASURED TOLERANCE
IMITS OF MAMMOGRAPHY x-RAY TUBES SPECIFIED BYNEMA

L
AND MQSA REGULATIONS
NOMINAL FOCAL SPOT SIZE (mm)
WIDTH (mm)
LENGTH (mm)
0.10
0.15
0.15
0.15
0.23
0.23
0.20
0.30
0.30
0.30
0.45
0.65
0.40
0.60
0.85
0.60
0.90
1.30
National Electrical Manufacturers Association publication XR-5, and MQSA regulations 21 CFR 900.
Focal spot resolvability with or without magnification is measured by imaging

a high-resolution bar pattern with up to 20 line pairs/mm. For conventional breast
imaging, the bar pattern is placed 4.5 cm above the breast support surface near the
chest wall; for magnification imaging, it is placed 4.5 cm above the magnification
platform. Orientation of the bar pattern parallel and perpendicular to the cathodeanode direction yields measurements of overall effective resolution that reflect the
effects of the focal spot dimensions, the acquisition geometry, and the detector sampling characteristics. A resolution bar pattern is shown in Figure 8-6B for a magnification study using the 0.1-mm focal spot. The resolving capability of the imaging
system is limited by the component that causes the most blurring. In magnification
mammography, this is generally the focal spot, whereas in contact mammography, it
may be the detector element size, and at other times patient motion.
Tube Port, Tube Filtration, and Beam Quality

The tube port and added tube filters play an important role in shaping the mammography x-ray energy spectrum. The tube port window is made of beryllium. The low
atomic number (Z=4) of beryllium and the small thickness of the window (0.5 to 1 mm)
allow the transmission of all but the lowest energy (less than 5 keV) bremsstrahlung
x-rays. In addition, Mo and Rh targets produce beneficial K-characteristic x-ray peaks at
17.5 and 19.6 keV (Mo) and 20.2 and 22.7 keV (Rh) (see Chapter 6, Table 6-2),
whereas tungsten targets produce a large fraction of unwanted L-characteristic x-rays
at 8 to 10 keV. Figure 8-7 shows a bremsstrahlung, characteristic, and composite x-ray
spectrum from an x-ray tube with a Mo target and Be window operated at 30kV.
Added x-ray tube filtration improves the energy distribution of the mammography
output spectrum by selectively removing the lowest and highest energy x-rays from
the x-ray beam, while largely transmitting desired x-ray energies. This is accomplished
by using elements with K-absorption edge (Chapter 3) energies between 20 and 27 keV.
Elements that have these K-shell binding energies include Mo, Rh, and Ag, and each
can be shaped into thin, uniform sheets to be used as added x-ray tube filters. At the
lowest x-ray energies, the attenuation of added filtration is very high. The attenuation decreases as the x-ray energy increases up to the K-edge of the filter element. For
x-ray energies just above this level, photoelectric absorption interactions dramatically
increase attenuation as a step or edge function (Fig8-8A). At higher x-ray energies,
the attenuation decreases. The result is the selective transmission of x-rays in a narrow band of energies from about 15 keV up to the K-absorption edge of the filter.
Molybdenum target
30 kVp
245
Bremsstrahlung
M
L
K
10
20
30
Energy (keV)
Characteristic
Binding Energy
K = 20 keV
L = 2.5 keV
M = 0.4 keV
10
20
Energy (keV)
K
e
30
Composite
high dose
low contrast
M K
19.6 keV
0
L K
17.5 keV
10
20
Energy (keV)
30
FIGURE 8-7 The x-ray spectrum of a mammography x-ray tube is composed of bremsstrahlung (with a
continuous photon energy spectrum) and characteristic (discrete energies) radiation (see Chapter 6 for more
details). A Mo anode tube operated at 30 kV creates the continuous spectrum as well as characteristic radiation with photon energies of 17.5 and 19.6 keV. On the lower right, the unfiltered composite spectrum
transmitted through 1 mm of Be (tube port material) has a large fraction of low-energy x-rays that will deposit
high dose without contributing to the image, and a substantial fraction of high-energy x-rays that will reduce
subject contrast of the breast tissues. The ideal spectral energy range is from about 15 to 25 keV, depending
on breast tissue composition and thickness.
In Figure 8-8B, the unfiltered Mo target spectrum and a superimposed attenuation

curve for a Mo filter are shown. Importantly, the characteristic x-ray energies produced
by the Mo target occur at the lowest attenuation of the filter in this energy range.
With a Mo target, a 0.030-mm-thick Mo filter or a 0.025-mm Rh filter is typically
used, and for a Rh target, a 0.025-mm Rh filter is used. A variety of filters are used
with W targets, including Rh (0.05 mm), Ag (0.05 mm), and Al (0.7 mm).
The spectral output of a Mo target and 0.030-mm-thick Mo filter is shown in
Figure 8-9A, illustrating the selective transmission of Mo characteristic radiation and
significant attenuation of the lowest and highest x-rays in the transmitted spectrum.
Tuning the spectrum to achieve optimal effective x-ray energy for breast imaging
is accomplished by selecting the anode material, added filtration material, and kV.
Screen-film detectors most often use a Mo target and 0.03-mm Mo filtration with a
kV of 24 to 25 kV for thin, fatty breasts and up to 30 kV for thick, glandular breasts.
For thicker and denser breasts, a Mo target and Rh filter are selected with higher
voltage, from 28 to 32 kV, to achieve a higher effective energy and more penetrating
beam (Fig. 8-9B, right graph). Some systems have Rh targets, which produce Rh
characteristic x-ray energies of 20.2 and 22.7 keV, achieving an even higher effective
energy x-ray beam for a set kV (Fig. 8-10A). A Mo filter should never be used with
a Rh target, because Rh characteristic x-rays are attenuated significantly as their
energies are above the Mo K-absorption edge (Fig. 8-10B).
246

A
Linear Attenuation Coefficient,

Mo, Rh, Ag, Al
30 kVp, Mo target, unfiltered

Relative number of x-rays
2000
(cm1)
1500
1000
Mo Rh Ag
Al
500
0
0
10
20
Energy, keV
30
40
10
20
30
40
Energy, keV
FIGURE 8-8 A. The linear attenuation coefficients of Al, Mo, Rh, and Ag are plotted as a function of energy.
A low-attenuation window exists below the K-absorption edge energy for the higher Z elements. B. An
unfiltered Mo target spectrum generated at 30 kV shows a large fraction of low- and high-energy photons.
Superimposed at the same energy scale is a dashed line illustrating a Mo filter attenuation as a function of
energy.
W targets are now used for many digital mammography systems because of their
higher bremsstrahlung production efficiency and higher tube loadings than Mo and
Rh targets. K-edge filters can optimize the output energy spectrum for breast imaging.
However, an unfiltered W spectrum contains a huge fraction of unwanted L x-rays in
the 8- to 12-keV range (Fig. 8-11A). Therefore, minimum filter thicknesses of 0.05mm
for Rh and Ag are needed to attenuate the L-x-rays to negligible levels, as shown in
Figure 8-11B. In some applications, an Al filter is used, but because of its low Z and
lack of a useful K-absorption edge, a thickness of 0.7mm is necessary to attenuate
the L x-rays. In digital tomosynthesis, in which contrast is largely provided by the
tomographic image reconstruction process (Section 8.5), breast dose can be reduced
by the use of Al filtration, which yields a larger fraction of high-energy x-rays.
A
30 kVp, Mo target, 0.025 mm Rh filter

(superimposed on 0.03 mm Mo filter spectrum)
30 kVp, Mo target, 0.03 mm Mo filter
10
20
Energy, (keV)
30
40
x-rays transmitted
using Rh filter
10
20
Energy, (keV)
30
40
FIGURE 8-9 A. A filtered output spectrum is shown for a Mo target and 0.030-mm Mo filter for a 30-kV
tube voltage, with prominent characteristic peaks at 17.5 and 19.6 keV. This is a common spectrum for less
dense and thinner breast tissues, particularly with lower operating tube voltages (25 to 28 kV). B. A filtered
output spectrum is shown for a Mo target and 0.025-mm Rh filter, superimposed on the Mo/Mo spectrum
in (A), indicating the transmission of higher energy bremsstrahlung x-rays up to the K absorption edge of Rh
at 23.2 keV. The spectrum, including the Mo characteristic x-rays, is more energetic and penetrating, and is
preferable for imaging thicker and denser breast tissues.
A
30 kVp, Rh target, 0.030 mm Mo filter
30 kVp, Rh target, 0.025 mm Rh filter
10
20
Energy, (keV)
30
40
247
Rh characteristic
x-rays absorbed
by Mo filter
10
20
Energy, (keV)
30
40
FIGURE 8-10 A. A filtered output spectrum is shown for a Rh target and 0.025-mm Rh filter for a 30-kV
tube voltage, with prominent Rh characteristic x-rays at 20.2 and 22.7 keV. This spectrum provides a higher
effective energy than a Mo targetRh filter x-ray beam at the same kV, due to the higher energy characteristic x-rays. B. A combination of a Rh target and Mo filter is inappropriate, as the Rh characteristic x-rays are
strongly attenuated as shown.
Half-Value Layer
The half-value layer (HVL) of a mammography x-ray beam ranges from 0.3 to 0.7-mm
Al for the kV range and combinations of target material, filter material, and filter
thickness used in mammography (Fig. 8-12). The HVL depends on the target material (Mo, Rh, W), kV, filter material, and filter thickness. Measurement of the HVL is
usually performed with the compression paddle in the beam, using 99.9% pure Al
sheets of 0.1-mm thickness. A Mo target, 0.03-mm Mo filter, 28 kV, and a 1.5 mm
Lexan compression paddle produce a HVL of about 0.35-mm Al, whereas a W target, 0.05-mm Rh filter, and 30 kV produce a HVL of about 0.55-mm Al. HVLs vary
30 kVp, W target: output normalized to 1 mGy

Filters: 0.05 mm Rh, 0.05 mm Ag, 0.7 mm Al
Rh
30 kVp, W target: unfiltered
L x-rays
Ag
Al
10
20
Energy, keV
30
40
10
20
Energy, keV
30
40
FIGURE 8-11 A. An unfiltered W target spectrum generated at 30 kV shows a large fraction of lower
energy L-Characteristic x-rays at 8 to 11.5 keV. Superimposed at the same energy scale is a dashed line illustrating the high- and low-attenuation characteristics of a Rh filter. B. Filtered W-target x-ray spectra by Rh
(0.05mm), Ag (0.05 mm), and Al (0.7 mm) are normalized to the same exposure output.
FIGURE 8-12 The HVL (including the

compression paddle attenuation) versus kV
is plotted for Mo targets with Mo and Rh
filters, and for W targets with Rh, Ag, and
Al filters. HVL measurements are representative of an average of a number of evaluations on various mammography systems at
the UC Davis Health System and are generally within / 0.05 mm of the values
shown.
Target / Filter HVL versus kV

plus 1.5 mm Lexan compression paddle
0.70
0.65
0.60
HVL (mm Al)
248
W / Ag
0.55
W / Rh
0.50
W / Al
0.45
Mo / Rh
0.40
Mo / Mo
0.35
0.30
20
25
30
kV
35
40
from machine to machine because of slight variation in actual filter thicknesses and
kV. The HVL of breast tissue is highly dependent on tissue composition (glandular,
fibrous, or fatty) and the HVL of the incident x-ray beam. Usually, the HVL for breast
tissues is from 1 to 3 cm.
Minimum HVL limits, listed in Table 8-2, are prescribed by MQSA regulations
to ensure that a beam of adequate effective energy is achieved. Maximum HVL limits are recommended by the ACR accreditation guidelines, as listed in Table 8-3.
An x-ray beam that is harder than optimal indicates too much filtration or a pitted anode or aged tube and can result in reduced output and poor image quality.
Estimates of the radiation dose to the breast require accurate assessment of the HVL
(see Section8.6).
Tube Output and Tube Output Rate

Tube output is a measure of the intensity of the x-ray beam, typically normalized to
mAs or to 100 mAs, at a specified distance from the source (focal spot). Common
units of tube output are mGy (air kerma)/100 mAs and mR (exposure)/mAs. The
kV, target, filter material and thickness, distance from the source, and focal spot size
must be specified. Figure 8-13 shows the output at a 50-cm distance from Mo and W
TABLE 8-2REQUIREMENTS FOR MINIMUM HVL IN THE MQSA
EGULATIONS (21 CFR PART 1020.30); ACR: WITH

R
COMPRESSION PADDLE
TUBE VOLTAGE (kV)
MQSA: kV/100
ACR: kV/100 0.03
24
0.24
0.27
26
0.26
0.29
28
0.28
0.31
30
0.30
0.33
32
0.32
0.35
ACR, American College of Radiology; CFR, Code of Federal Regulations; MQSA, Mammography Quality Standards Act.
249
TABLE 8-3ACR SPECIFICATIONS FOR MAXIMUM HVL: MAXIMUM HVL
(mm AI) = kV/100 C FOR T

ARGET-FILTER COMBINATIONS
TUBE VOLTAGE (kV)
Mo/Mo C = 0.12
Mo/Rh C = 0.19
Rh/Rh C = 0.22
W/Rh C = 0.30
24
0.36
0.43
0.46
0.54
26
0.38
0.45
0.48
0.56
28
0.40
0.47
0.50
0.58
30
0.42
0.49
0.52
0.60
32
0.44
0.51
0.54
0.62
ACR Accreditation Program requirements as specified in the Medical Physics testing section, http://www.acr.org
target x-ray tubes with a variety of tube filter materials and thicknesses. Even though
W targets are more efficient at producing x-rays, the thicker filters needed to attenuate the L-characteristic x-rays result in lower tube output per mAs compared to the
Mo target. However, W spectra have higher HVLs and greater beam penetrability,
allow higher tube current, and result in comparable exposure times to a Mo target
and filter for a similar breast thickness.
x-ray tube output values are useful for calculating the free-in-air incident air
kerma (or exposure) to the breast for a mammography systems target and filter
combination, kV, mAs, and source-to-breast surface distance. The source-to-breast
surface distance is determined from the known SID, breast platform to detector distance, and compressed breast thickness. For instance, assume that a mammography
system with a Mo target and Rh filter uses 30 kV and 160 mAs for a SID of 65 cm,
compressed breast thickness of 6 cm, and a breast platform to detector distance of
2cm. The entrant breast surface to the source is closer by 8 cm, and is therefore
57cm from the source. From Figure 8-13, the tube output is 16 mGy/100 mAs for
30kV at a distance of 50 cm. Calculation of incident air kerma considers tube output
at a specific kV, the mAs used, and inverse square law correction from 50 to 57 cm:
16 mGy/100 mAs 160 mAs [50.0/57.0]2 19.7 mGy
Calculation of the average glandular dose to the breast is determined from the
measured incident air kerma value and other parameters as discussed in Section 8.6.
FIGURE 8-13 Tube output (mGy/100mAs
at 50-cm distance from the source with
compression paddle in the beam) for two
clinical mammography units is measured at
2-kV intervals, and the data are smoothed.
A Mo target with 0.030-mm Mo and
0.025-mm Rh filter, and a W target with
0.05-mm Rh, 0.05-mm Ag, and 0.7-mm Al
filters are plotted.
250
Tube output rate is the air kerma rate at a specified distance from the x-ray focal
spot and is a function of the tube current achievable for an extended exposure time
(typically ~300 mAs for an exposure time greater than 3 s). To ensure the ability to
deliver a sufficient x-ray beam fluence rate to keep exposure times reasonable, MQSA
regulations require that systems be capable of producing an air kerma rate of at least
7.0mGy/s, when operating at 28 kV in the standard (Mo/Mo) mammography mode,
at any SID for which the system is designed to operate for screen-film detectors. Digital systems have requirements specified by the manufacturers QC manuals.
x-ray Tube Alignment and Beam Collimation

Alignment of the x-ray tube and collimator are crucial to ensure that the x-ray beam
central axis is perpendicular to the plane of the image receptor and intercepts the
center of the chest wall edge of the image receptor. This will protect the patient from
unnecessary radiation exposure to the lungs and torso and include as much breast tissue as possible in the image. Tube alignment with respect to the image receptor must be
verified during acceptance testing of the system and after x-ray tubereplacements.
Collimation of the x-ray beam is achieved by the use of fixed-size metal apertures
or adjustable shutters. For most screen-film examinations, the field size is automatically
set to match the film cassette dimensions (e.g., 18 24 cm or 24 30 cm). For a large
area digital detector (24 30 cm), when operating with the smaller field area (18
24cm), one of three active acquisition areas is used: center, left shift, and right shift,
which requires the collimator assembly to restrict the x-ray field to the corresponding
active detector area. Field shifts are used for optimizing the oblique projections for subjects with smaller breasts to accommodate positioning of the arm and shoulder at the top
edge of the receptor. Collimation to the full active detector area is the standard of practice
(see Section 8.4: Film-Viewing Conditions). There is no disadvantage to full-field collimation compared to collimation to the breast only, as the tissues are fully in the beam in
either case. However, for magnification and spot compression studies, manually adjusted
shutters allow the x-ray field to be more closely matched to the volume being imaged.
The projected x-ray field must extend to the chest wall edge of the image receptor without cutoff, but not beyond the receptor by more than 2% of the SID. If the
image shows evidence of collimation of the x-ray field on the chest wall side or if
the chest wall edge of the compression paddle is visible in the image, service must
be requested. A collimator light and mirror assembly visibly display the x-ray beam
area. Between the tube port and the collimator is a low-attenuation mirror that directs
light from the collimator lamp through the collimator opening to emulate the x-ray
field area. Similar to conventional x-ray tube collimator assemblies, the light field
must be congruent with the actual x-ray field to within 2% overall, which is achieved
by adjustment of the light and mirror positions. MQSA regulations require repairs to
be made within 30 days if the light/x-ray field congruence, x-ray field-image receptor
alignment, or compression paddle alignment is out of tolerance as described above.
8.2
x-ray Generator and Phototimer System

x-ray Generator
A dedicated mammography x-ray generator is similar to a conventional x-ray generator in design and function. Differences include the lower voltages supplied to
the x-ray tube, space charge compensation, and automatic exposure control (AEC)
251
c ircuitry. In most cases, AEC is employed for mammography screening and diagnostic
examinations, although there are instances when the technologist will use manual
controls to set the tube current - exposure duration product (mAs). Like most contemporary x-ray imaging systems, high-frequency generators are used for mammography due to low voltage ripple, fast response, easy calibration, long-term stability,
and compact size. There are circuits that prohibit x-ray exposures if the setup is
incomplete (e.g., insufficient compression, cassette not in the tunnel, or digital detector not ready to capture an image).
Automatic Exposure Control

The AEC employs a radiation sensor or sensors, a charge amplifier, and a voltage comparator to control the exposure (Fig. 8-14). For cassette-based image receptors (screenfilm and computed radiography [CR]), the phototimer sensor is located underneath the
cassette, and consists of a single ionization chamber or an array of three or more semiconductor diodes. x-rays transmitted through the breast, antiscatter grid (if present),
and the image receptor generate a signal in the detector. The signal is accumulated
(integrated) and, when the accumulated signal reaches a preset value, the exposure
is terminated. The preset value corresponds to a specified signal-to-noise ratio (SNR)
in a digital mammography unit or an acceptable optical density (OD) if a film-screen
system is used. For an active matrix flat-panel imager, the detector array itself can be
used to measure the x-rays transmitted through the breast and the antiscatter grid.
Phototimer algorithms use several inputs to determine the radiographic technique, including compressed breast thickness, phototimer adjustment settings on
the generator console, the kV, the tube anode selection (if available), and the tube
filter to achieve the desired film OD or digital SNR in the acquired image. The operator typically has two or three options for AEC: (a) a fully automatic AEC mode
that sets the optimal kV and filtration (and target material on some systems) from a
short test exposure of approximately 100 ms to determine the penetrability of the
breast; (b) automatic kV with a short test exposure, with user-selected target and
filter values; and (c) automatic time of exposure using manually set target, filter, and
High
Frequency
Generator
Auto mode
AEC Logic
kV-filter
Thickness
Penetrability
SNR / Density
selector
1 0 +1
2
+2
3
+3
4
+4
+5
5
Manual
mAs
kV
AEC
Auto kV-filter
Auto kV
Auto time
Reference Voltage
Backup Timer
Termination
Circuit
Amplifier
Capacitor
Voltage
Comparator
Phototimer sensor
FIGURE 8-14 The AEC (phototimer) circuits use algorithms with input from the breast thickness (from
the compression paddle), breast density (via a test exposure), density setting (operator density selector), and
exposure duration (reciprocity law failure issues) to determine overall exposure time. In fully automatic modes,
the kV, beam filter, and target are also determined, usually from a short (100 ms) exposure to determine
attenuation characteristics of the breast.
252
kV values. For most patient imaging, the fully automatic mode is used, but, for QC
procedures, the automatic exposure time adjustment mode with other parameters
manually selected by the user is often employed.
At the generator control panel, an exposure adjustment selector modifies the
AEC response to permit adjustments for unusual imaging circumstances such as
imaging breasts with implants, magnification, or for radiologist preference in film
OD or digital image SNR. Usually, eleven steps (5 to 0 [neutral] to 5) are available
to increase or decrease exposure by adjusting the AEC comparator switch. Each step
provides a difference of 10% to 15% positive (greater exposure) or negative (lesser
exposure) per step from the baseline (0) setting.
An x-ray exposure that is not stopped by the AEC circuit and exceeds a preset
time (e.g., greater than 5 s) is terminated by a backup timer. This can occur if there
is a malfunction of the AEC system or if the kV is set too low with insufficient x-ray
transmission. In the latter situation, the operator must select a higher kV to achieve
greater beam penetrability and shorter exposure time to correct the problem.
Inaccurate phototimer response, resulting in an under- or overexposed film or
digital image, can be caused by breast tissue composition (adipose versus glandular) heterogeneity, compressed thickness beyond the calibration range (too thin or
too thick), a defective cassette, a faulty phototimer detector, or an inappropriate
kV setting. Screen-film response to very long exposure times, chiefly in magnification studies caused by the low mA capacity of the small focal spot, results in
reciprocity law failure (Chapter 7), whereby the resultant OD is less than would be
predicted by the amount of radiation delivered to the screen-film cassette; consequently, insufficient film OD results, and extension of the exposure time is added
to compensate. For extremely thin or fatty breasts, the phototimer circuit and x-ray
generator can be too slow in terminating the exposure, causing film overexposure.
In situations where the exposure is terminated too quickly, grid-line artifacts will
commonly appear in the image due to lack of complete grid motion during the
short exposure.
The position of the phototimer detector (e.g., under dense or fatty breast tissue)
can have a significant effect on film density but has much less of an impact on digital
images because of postprocessing capabilities, even though the SNR will vary. Previous mammograms can aid the technologist in selecting the proper position for the
phototimer detector to achieve optimal film density or optimal SNR for glandular
areas of the breast. Most systems allow phototimer positioning in the chest wall to
anterior direction, while some newer systems also allow side-to-side positioning to
provide flexibility for unusual circumstances.
Technique Chart
Technique charts are useful guides to help determine the appropriate kV and targetfilter combinations for specific imaging tasks, based on breast thickness and breast
composition (fatty versus glandular tissue fractions). Most mammographic techniques use phototiming, and the proper choice of kV is essential for a reasonable
exposure time, defined as a range from approximately 0.5 to 2.0 s to achieve an
OD of 1.5 to 2.0 for film, or the desired SNR for digital images. Too short an exposure can cause visible grid lines to appear on the image, whereas too long an exposure can result in breast motion, either of which degrades the quality of the image.
Table 8-4 lists kV recommendations for screen-film detectors to meet the desired
exposure time range for breast thickness and breast composition using a Mo target
and 0.030-mm Mo filter, determined from computer simulations and experimental
253
TABLE 8-4RECOMMENDED kV, AS A FUNCTION OF BREAST COMPOSITION
AND THICKNESS
BREAST THICKNESS (cm)
BREAST COMPOSITION
Fatty
24
24
24
24
25
27
30
50/50
24
24
24
25
28
30
32
Glandular
24
24
26
28
31
33
35
The goal is an exposure time between 0.5 and 2.0 s. This chart is for a Mo target and 0.030-mm Mo filter using
a 100-mA tube current and screen-film detector. These kVs may not be appropriate for other target/filter
combinations or for digital detectors. The kV values were determined by computer simulations and clinical
measurements.
easurements. Techniques for digital detector systems are more flexible, and
m
because of postprocessing and contrast enhancement capabilities, will often use a
wider range of kV settings. Atechnique chart example for an amorphous selenium
(Se) flat panel detector system using a Mo target, a W target, or a W target (without
a grid for a digital tomosynthesis acquisition) and corresponding tube filtration is
listed in Table8-5.
8.3
Compression, Scattered Radiation, and Magnification

Compression
Breast compression is an important part of the mammography examination, whether
using screen-film or digital detectors. Firm compression reduces overlapping
anatomy, decreases tissue thickness, and reduces inadvertent motion of the breast
(Fig.8-15A). It results in fewer scattered x-rays, less geometric blurring of anatomic
structures, and lower radiation dose to the breast tissues. Achieving a uniform breast
thickness lessens exposure dynamic range and allows the use of higher contrast film,
or allows more flexibility in image processing enhancement of the digital image.
Compression is achieved with a compression paddle, a Lexan plate attached to
a mechanical assembly (Fig. 8-15B). The full area compression paddle matches the
size of the image receptor (18 24 cm or 24 30 cm) and is flat and parallel to the
breast support table. An alternative to the flat compression paddle is the flex paddle
TABLE 8-5TECHNIQUE CHART FOR A DIGITAL MAMMOGRAPHY S

YSTEM
WITH SELENIUM DETECTOR FOR 50% GLANDULAR 50%

ADIPOSE COMPOSITION
DIGITAL, Mo TARGET,
CELLULAR GRID
DIGITAL, W TARGET,
CELLULAR GRID
DIGITAL TOMOSYNTHESIS,
W TARGET, NO GRID
Thickness (cm)
Filter
kV
mAs
Filter
kV
mAs
Filter
kV
mAs
<3
Mo 30 mm
24
35
Rh 50 mm
25
40
Al 700 mm
26
35
35
Mo 30 mm
27
75
Rh 50 mm
28
80
Al 700 mm
29
50
57
Mo 30 mm
31
100
Rh 50 mm
31
150
Al 700 mm
33
65
>7
Rh 25 mm
33
150
Ag 50 mm
32
200
Al 700 mm
38
85
254
FIGURE 8-15 A. Compression is

essential for mammographic studies to reduce breast thickness (less
scatter, reduced radiation dose,
and shorter exposure time) and to
spread out superimposed anatomy.
B.Suspicious areas often require
spot compression to eliminate
superimposed anatomy by further
spreading the breast tissues over a
localized area. C. Example of image
with suspicious finding (left). Corresponding spot compression image
shows no visible mass or architectural distortion (right).
that is spring loaded on the anterior side to tilt and accommodate variations in breast
thickness from the chest wall to the nipple, providing more uniform compression of
the breast. A compression paddle has a right-angle edge at the chest wall to produce
a flat, uniform breast thickness when an adequate force of 111 to 200 newtons (25
to 44 lb) is applied. A smaller spot compression paddle (~5-cm diameter) reduces
the breast thickness further over a specific region and redistributes the tissue for
improved contrast and reduced anatomic overlap (Fig.8-15C,D).
255
Typically, a hands-free (e.g., foot-pedal operated), motor-driven compression

paddle is used, which is operable from both sides of the patient. In addition, a
mechanical adjustment knob near the paddle holder allows fine manual adjustments
of compression. While firm compression is not comfortable for the patient, it is often
necessary for a clinically acceptable image.
Scattered Radiation and Antiscatter Grids

x-rays transmitted through the breast contain primary and scattered radiation.
Primary radiation carries information regarding the attenuation characteristics of the
breast and delivers the maximum possible subject contrast to the detector. Scattered
radiation is an additive, gradually varying radiation distribution that degrades subject contrast and adds random noise. If the maximum subject contrast without scatter is C0 and the maximum contrast with scatter is CS, then the contrast degradation
factor (CDF) is a pproximated as
CDF =
Cs
1
=
C 0 1 + SPR
[8-2]
where SPR is the scatter-to-primary ratio (Chapter 7). x-ray scatter increases with
increasing breast thickness and breast area, with typical SPR values shown in
Figure8-16. A breast of 6-cm compressed thickness will have an SPR of approximately
0.6 and a calculated scatter degradation factor of (1/(10.6)) 0.625 62.5%.
Without some form of scatter rejection, therefore, only a fraction of the inherent
subject contrast can be detected.
In digital mammography, unlike screen-film mammography, the main adverse
effect of scattered x-rays is not a reduction of contrast. Contrast can be improved by
any desired amount by post-acquisition processing, such as windowing. The main
adverse effect of scatter in digital mammography is that scattered photons add random noise, degrading the signal to noise ratio.
FIGURE 8-16 x-ray scatter redu
ces the radiographic contrast of
the breast image. Scatter is chiefly
dependent on breast thickness and
field area, and largely independent
of kV in the mammography energy
range (25 to 35 kV). The scatter-toprimary ratio is plotted as a function
of the diameter of a semicircular
field area aligned to the chest wall
edge, for several breast thicknesses
of 50% glandular tissue.
Mo/Mo 50% glandular tissue

1.2
1.1
8 cm
Scatter to primary ratio
1.0
0.9
0.8
0.7
6 cm
0.6
0.5
4 cm
0.4
0.3
0.2
2 cm
0.1
0.0
10
15
20
Diameter (cm)
25
30
256
Scattered radiation reaching the image receptor can be greatly reduced by the
use of an antiscatter grid or air gap. (Antiscatter grids are described in Chapter 7.)
For contact mammography, an antiscatter grid is located between the breast and the
detector. Mammography grids transmit about 60% to 70% of primary x-rays and
absorb 75% to 85% of the scattered radiation. Linear focused grids with grid ratios
(height of the lead septa divided by the interspace distance) of 4:1 to 5:1 are common
(e.g., 1.5 mm height, 0.30-mm distance between septa, 0.016-mm septa thickness),
with carbon fiber interspace materials (Fig. 8-17A). Grid frequencies (number of
lead septa per cm) of 30/cm to 45/cm are typical. To avoid grid-line artifacts, the grid
must oscillate over a distance of approximately 20 lines during the exposure. Excessively short exposures are the cause of most grid-line artifacts because of insufficient
grid motion.
A cellular grid, made of thin copper septa, provides scatter rejection in two
dimensions (Fig. 8-17B). Specifications of this design include a septal height of
2.4mm, 0.64-mm distance between septa (3.8 ratio), a septal thickness of 0.03 mm,
and 15cells/cm. During image acquisition, a specific grid motion is necessary for
complete blurring, so specific exposure time increments are necessary and are determined by AEC logic evaluation of the first 100 ms of the exposure. Because of two
dimensional scatter rejection and air interspaces, the cellular grid provides a better
contrast improvement factor than the linear grid.
Grids impose a dose penalty to compensate for loss of primary radiation that
would otherwise contribute to SNR in the digital image or for x-ray scatter and primary radiation losses that would otherwise contribute to film optical density in filmscreen mammography. As discussed in Chapter 7, the Bucky factor is the ratio of
exposure with the grid compared to the exposure without the grid to achieve the
same film optical density. For mammography grids, the Bucky factor is 2 to 3, so the
breast dose is doubled or tripled, but the benefit is improvement of image contrast
by up to 40%. For digital acquisitions, there is no similar Bucky factor definition,
but the loss of signal in the digital image from primary radiation being attenuated by
the grid strips will increase the exposure needed to achieve a similar SNR. Typically,
Linear grid structure:

1D scatter rejection
Cellular grid structure:

2D scatter rejection
Air gap and magnification:

geometric scatter rejection
FIGURE 8-17 Antiscatter devices commonly employed in mammography include (A) the linear grid of
approximately 5:1 grid ratio and carbon fiber interspace material, (B) a cellular crosshatch grid structure made
of copper sheet of approximately 3.8 grid ratio with air interspaces and scatter rejection in two dimensions,
and (C) the air gap intrinsic to the magnification procedure. Note: while the illustration depicts 100% scatter
rejection by the grid, approximately 15% of scattered x-rays are transmitted.
257
the extra radiation needed for a digital detector is less than the Bucky factor for
screen-film, perhaps by one half or more. Grid performance is far from ideal, but
digital acquisition and image processing can somewhat mitigate the dose penalty
incurred by the use of the antiscatter grid and analog screen-film detectors.
An air gap reduces scatter by increasing the distance of the breast from the detector, so that a large fraction of scattered radiation misses the detector (Fig.8-17C).
The consequences of using an air gap, however, are that the field of view (FOV) is
reduced, the magnification of the breast is increased, and the dose to the breast is
increased. However, use of a sufficiently large air gap can render the antiscatter grid
unnecessary, thus reducing dose to the breast by approximately the same factor.
Magnification Techniques
Geometric magnification (Chapter 7) is used to improve system resolution, typically for better visualization of microcalcifications. In magnification studies, geometric magnification is achieved by placing a breast support platform (a magnification
stand) at a fixed distance above the detector, selecting the small (0.1 mm) focal
spot, removing the antiscatter grid, and using a compression paddle (Fig. 8-18).
Most dedicated mammographic units offer magnifications of 1.5, 1.8, or 2.0.
Advantages of magnification include (a) increased effective resolution of the image
receptor by the magnification factor; (b) reduction of image noise relative to the
objects being rendered; and (c) reduction of scattered radiation. Geometric magnification of the x-ray distribution pattern relative to the inherent unsharpness of the
image receptor improves effective resolution, as long as the loss of resolution by geometric blurring is mitigated with the small 0.1-mm nominal focal spot. Since there
are more x-rays per object area in a magnified image, the effective quantum noise is
reduced, compared to the standard contact image. The distance between the breast
support surface of the magnification stand and the image receptor reduces scattered
radiation reaching the image receptor, thereby improving image contrast and reducing random noise and making an antiscatter grid unnecessary.
Magnification has several limitations, the most significant being geometric
blurring caused by the finite focal spot area. Even with a small 0.1-mm focal spot,
spatial resolution will be less on the cathode side of the x-ray field (toward the chest
FIGURE 8-18 Geometric magnification. A support platform positions the
breast closer to the source focal spot,
resulting in 1.5 to 2.0 image magnification. A small focal spot (0.1-mm
nominal size) reduces geometric blurring.
Variation of effective focal spot size along
the cathode-anode axis (large to small
effective size) is accentuated with magnification and increased penumbra width.
Best resolution and detail in the image
exist on the anode side of the field. SID:
source to image distance; SOD: source to
object (midplane) distance; OID: object to
image distance.
258
wall), where the effective focal spot length is largest. In addition, the small focal
spot has a tube current limit of about 25 mA for a Mo target, so a 100-mAs technique requires 4 s exposure time. Even slight breast motion will result in image
blurring, which is exacerbated by the magnification. For screen-film detectors and
long exposure times, reciprocity law failure requires additional radiation exposure
to achieve the desired film OD. Regarding breast dose, the elimination of the grid
reduces dose by a factor of 2 to 3, but the shorter distance between the x-ray focal
spot and the breast increases dose by about the same factor due to the inverse square
law. Therefore, in general, the average glandular dose delivered to the breast with
magnification is similar to that from contact mammography. However, the smaller
irradiated FOV justifies collimating the x-ray beam to only the volume of interest,
which reduces radiation scatter and breast dose.
8.4
Screen-Film Cassettes and Film Processing

Screen-film detectors in mammography have been a proven technology over the last
30 years; in fact, the 20012005 study comparing screen-film and digital mammography detectors showed that screen-film detectors are comparable to digital detectors in terms of diagnostic accuracy as a means of screening for breast cancer. Digital
mammography has been shown to be more accurate for women under the age of
50years, women with radiographically dense breasts, and premenopausal or perimenopausal women compared to screen-film mammography (Pisano etal. 2005).
While digital mammography has reduced the use of screen-film mammography, for
some low-volume mammography clinics and those clinics not yet able to transition
to digital technology, screen-film detectors remain a reasonable and cost-effective
technology.
Design of Mammographic Screen-Film Systems

A mammographic screen-film detector is comprised of a cassette, intensifying screen
or screens, and a light-sensitive film. Most cassettes are made of low-attenuation carbon fiber and have a single high-definition phosphor screen used in conjunction with
a single emulsion film. Terbium-activated gadolinium oxysulfide (Gd2O2S:Tb) is the
most commonly used screen phosphor. This scintillator emits green light, requiring a
green-sensitive film emulsion. Intensifying screen-film speeds and spatial resolution
are determined by screen phosphor particle size, number of phosphor particles per
volume, light-absorbing dyes in the phosphor matrix, and phosphor layer thickness.
Mammography screen-film systems have their own speed ratings and are classified as
regular (100 or par speed) and medium (150 to 190 speed), whereby the 100-speed
screen-film system requires 120 to 150 Gy incident air kerma to achieve the desired
film OD. For comparison, a conventional 400-speed screen-film cassette for general diagnostic imaging requires about 5-Gy air kerma.
The intensifying screen is positioned in the back of the cassette so that x-rays
travel through the cassette cover and the film before interacting with the phosphor.
Exponential attenuation in the phosphor results in a greater fraction of x-ray interactions near the phosphor surface closest to the film emulsion. At shallow depths in
the phosphor, absorbed x-rays produce light distributions that have minimal light
spread prior to interacting with the film, thus preserving spatial resolution. x-ray
absorption occurring at greater depth in the screen produces a broader light distribution and reduces spatial resolution (Fig. 8-19), but the number of interactions
259
FIGURE 8-19 Nearly all screen-film systems

used in mammography employ a single phosphor screen with a single emulsion film and
are front-loaded (x-rays pass through the
film to the screen). Light spread varies with
the depth of x-ray absorption, as depicted in
the lower diagram. Most x-rays interact in the
layers of the screen closest to the film, preserving spatial resolution because of reduced
spread of the light distribution.
decreases exponentially with depth. For a 100-speed mammography screen-film

cassette, the limiting spatial resolution is 15 to 20 lp/mm.
Selection of a screen-film combination requires consideration of measures such as
spatial resolution and radiographic speed, as well as factors such as screen longevity,
cassette design, film base color, film contrast characteristics, radiologists preference,
and cost. In todays market, there is no single best choice; hence, most screen-film
receptors, when used as recommended with particular attention to film processing,
provide excellent image quality for mammography imaging.
Film Recording and Film Processing

Mammography films are sensitized by x-rayinduced light from the phosphor screen.
Light produces latent image centers on microscopic light-sensitive silver halide grains in
the emulsion layered on a film substrate. (See Chapter 7 for more information.) Intensity variations in the light produce corresponding variations in the number of sensitized
grains per area. Chemical processing in a bath of developer solution renders the invisible
latent image into a visible image by reducing the sensitized grains to elemental silver.
Unsensitized grains in the film emulsion are not reduced and are dissolved and washed
away by the fixer solution. Subsequent rinsing and drying complete the film processing. Film is thus a two-dimensional recording of the incident light patterns encoded by
silver grains on the substrate that produce an image with varying light intensities when
the film is placed on a lightbox. Film opacity is the degree of the obstruction of light
transmission (the inverse of transmittance), and optical density (OD), equal to the base
ten logarithm of the opacity, is used as the metric of film response, as the eye responds
linearly to variations in OD (e.g., an OD = 2 appears to be twice as dark as an OD = 1,
although 10 times more light is attenuated with an OD = 2 compared to OD = 1). The
useful OD levels on a mammography film range from about 0.2 up to about 3.
Film processing is a critical step in the mammographic imaging chain for screenfilm detectors and is performed by automatic film processors. Film characteristics
such as film speed (an indication of the amount of incident radiation required to
achieve a specified OD on the film) and film contrast (the rate of change of OD
for a known difference in incident radiation) are consistently achieved by following
the manufacturers recommendations for developer formulation, development time,
260
developer temperature, developer replenishment rate, and processor maintenance. If

the developer temperature is too low, the film speed and film contrast will be reduced,
requiring a compensatory increase in radiation dose. If the developer temperature is
too high or immersion time too long, an increase in film speed occurs, permitting a
lower dose; however, the film contrast is likely to be reduced because film fog and
quantum mottle are increased. Stability of the developer may also be compromised at
higher-than-recommended temperatures. The manufacturers guidelines for optimal
processor settings and chemistry specifications should be followed. Also, because
mammographic films must be retained for years, fixer retention on the film must be
measured when evaluating the processor. If film washing is incomplete and fixer is
retained, the films will oxidize and image quality will deteriorate over time.
Film Response, Sensitometry, and Quality Control

A film characteristic curve, also referred to as an H and D curve, represents the relationship between the incident x-ray exposure (proportional to light intensity on the
film) and the resulting OD in the processed film. As shown in Figure 8-20A, a film
characteristic curve is a graph of the OD as a function of the logarithm of the relative exposure and has three major sections: (a) the toe has a low OD that changes
very little with incident exposure up to a threshold; (b) the linear section represents
increasing OD with increasing incident exposure, with subject contrast mapped as
optical densities that render the radiographic contrast; and (c) the shoulder represents saturation of the film, where increases in incident exposure do not significantly
increase the OD. A films characteristic curve is determined by both the design of
the film and the film processing. Another way of displaying the films response is the
gradient curve, shown in Figure 8-20B. The gradient curve is a graph of the change
of OD per unit change in the incident exposure, as a function of the logarithm of the
relative exposure. The gradient is the rate of change of the characteristic curve, that
is, the film contrast. The Film contrast changes with incident exposure and has a
value near the maximum only over a narrow range of incident exposures.
Because of the high sensitivity of mammography film quality to slight changes
in processor performance, routine monitoring of film contrast, speed, and base
plus fog OD is important. A film-processor QC program is required by the MQSA
A
4.5
Mammography Film Gradient

4.5
Shoulder
4.0
4.0
3.5
3.5
3.0
Linear
OD change
Optical Density
Mammography Characteristic Curve
2.5
2.0
1.5 1.0 OD+B&F
1.0
0.5
Exposure
latitude
3.0
2.5
2.0
1.5
1.0
0.5
Toe
0.0
0.0
10
100
Incident air Kerma (Gy)
1000
10
100
1000
Incident air Kerma (Gy)
FIGURE 8-20 A. The film characteristic curve is a graph of the OD as a function of the logarithm of incident
air kerma, mGy (or exposure, mR). B. The gradient curve (the rate of change of the characteristic curve) indicates the contrast as a function of the logarithm of air kerma (exposure). The maximal value of the gradient
curve is achieved for an exposure of about 120 mGy (14 mR), which is OD of about 2.0 for this exposure. Thus,
the maximal contrast (change in OD for a change in exposure) is achieved at an OD of about 2.0.
261
regulations, and daily sensitometric tests must be performed before the first clinical
images to verify acceptable performance. This requires the use of a sensitometer
and a densitometer. A sensitometer is a device that emulates a range of incident
radiation exposures by using a constant light source and calibrated optical attenuation steps to expose a mammographic film to known relative light intensities.
Developing the film produces a range of numbered OD steps known as a film
sensitometric strip. A densitometer is a device that measures the light transmission
through a small area of the film and calculates the OD. Additionally, a thermometer
is used to measure the temperature of the developer solution in the processor and
monitoring chart are tools used to complete the processor evaluation. A chart is
used to record and monitor the quality control measurements.
A box of film is reserved for QC testing; using films from a single box eliminates
variations among different lots of film from the manufacturer. Daily, in the darkroom
prior to the first case of the day, a film is exposed using a calibrated sensitometer
and is then developed (Fig. 8-21A). On the processed film, optical densities on the
sensometric strip are measured with a film densitometer (Fig.8-21B). Data, which
are plotted on a processor QC chart, are the base plus fog OD (measured at an area
on the film that was not subjected to light exposure), a mid-density speed index step,
and a density difference (an index of contrast) between two steps. The mid-density
OD step and the two steps for calculating the density difference are selected from
5day average ODs when the QC program is established. The mid-density step is the
one whose OD is closest to, but does not exceed 1.2; the density difference steps are
those whose ODs are closest to but not exceeding 0.45 and closest to 2.20. Action
limits, determined from the initial 5-day averages, define the range of acceptable
processor performance; they are base fog density not exceeding 0.03, middensity
0.15 OD, and density difference 0.15 OD. If the test results fall outside of these
limits, c orrective action must be taken prior to patient imaging.
A
Sensitometer
Film processor
Exposed film
Developed film
sensitometry strip
Exposure button
Unexposed Film
Calibrated light filter
Fixed intensity
light source
Density
Difference
Mid-density
Level
Densitometer
Optical
Density
Sensitometry Photocell
strip
Fixer Wash Dry
Developer
Exposure
timer
1.39
Calibrated
light source
1.30
1.95
1.25
1.90
1.20
1.85
1.15
1.80
1.10
1.75
Aperture
5
Action Limits
10
Day of Month
10
Day of Month
FIGURE 8-21 A. Film-processor testing requires a calibrated sensitometer to flash an unexposed film to a
specified range of light intensities using an optical step wedge. The film is then developed. B. The corresponding optical densities on the processed film are measured with a densitometer, and Base Fog (the lowest OD
value), the mid-density level step (a measurement indicating relative speed), and the density difference value,
calculated by the subtraction of a low OD step from a high OD step (indicating the relative contrast response),
are plotted to verify processor performance before patients are imaged.
262

4.5
Low
contrast
4.0
Radiographic contrast
Skin line
Optical Density
3.5
3.0
Adipose
High
contrast
2.5
Adipose /
Glandular
2.0
1.5
1.0
Dense
Glandular
Low
contrast
0.5
Subject contrast
0.0
10
30
100
300
1000
Incident air kerma(Gy)

FIGURE 8-22 Equal x-ray subject contrast (range of transmitted relative exposures) generated in the breast
are manifested with variable contrast in the film image, resulting from the nonlinear characteristic curve.
Areas in the breast that have high transmitted exposure, such as the skin line, and low transmitted exposure,
such as glandular tissue, will have reduced radiographic contrast. Only over a small range of exposures is the
radiographic contrast high.
Variability of Screen-Film Contrast Response

Radiographic contrast (i.e., OD differences on the developed film) results from variations in subject contrast (x-ray transmission differences due to anatomy) translated
into these OD differences according to the films characteristic curve. Therefore, the
radiographic contrast varies depending upon the incident exposure to each area on
the film-screen cassette. The best recording of subject contrast occurs on areas of the
film where the exposure corresponds to the steepest part of the film characteristic
curve. Significant reduction of contrast occurs on parts of the film where the exposures correspond to the toe and shoulder segments of the curve. Unfortunately, not
all important anatomical information is rendered with optimal contrast, particularly
in dense glandular breast tissues and at the skin line, as illustrated in Figure 8-22.
Film Viewing Conditions

Viewing conditions are very important so that the information on the film is visualized well by the radiologist. Subtle lesions on the mammographic film may be
missed if viewing conditions are suboptimal. Since mammography films are exposed
to high optical densities to achieve high contrast, view boxes in the mammography reading area must provide high luminance (a measure of brightness as described
in Chapter5). Mammography view boxes should have minimum luminances of at
least 3,000cd/m2, and luminances exceeding 6,000 cd/m2 are common. (Note: the
luminance of 1 cd/m2 is approximately the luminous intensity of the sky at the western horizon on a clear day, 15 min after sunset.) In comparison, the luminance of a
view box in diagnostic radiology is typically about 1,500 cd/m2.
A high-intensity bright light should be available in the reading room to penetrate high OD regions of the film, such as the skin line and the nipple area. Furthermore, the use of a magnifying glass improves the visibility of fine detail in the image,
such as microcalcifications.
Film masking (blocking clear portions of the film and areas of the view box
that are not covered by film to prevent extremely bright light from degrading the
Collimated to Breast
263
Collimated to Film
FIGURE 8-23 Optimal viewing of film mammograms requires that the entire area of the film be exposed to
radiation to darken the background surrounding the breast image; otherwise, light cannot easily be masked
when viewing the films on a viewbox. For digital acquisitions, the background area location in each image is
determined, and pixel padding is performed so that the background is dark, even when digitally inverting
contrast.
r adiologists visual adaptation to the range of luminances displayed by the images)

is essential for preserving perceived radiographic contrast. The film must be fully
darkened in background areas around the breast (Fig. 8-23) by collimating the x-ray
beam to the full area of the detector (see Section 8.1: Collimation). Most dedicated
mammography view boxes have adjustable shutters for masking to the film size.
The ambient light intensity (as measured by the illuminance) in a mammography
reading room must be reduced to eliminate reflections from the film and to improve
perceived radiographic contrast. Illuminance is the luminous flux incident upon a
surface per unit area, measured in lux or lumens/m2. For example, the full moon
in a clear night sky produces about 1 lux, whereas normal room lighting generates
100 to 1,000 lux. Subdued room lighting providing an illuminance of about 20-40
lux is acceptable for the viewing and interpretation of mammograms.
8.5
Digital Mammography
Digital acquisition devices became available in mammography in the early 1990s
in the form of small field-of-view digital biopsy systems. In early 2000, a full-field
digital mammography system was first approved by the FDA; it had a thin-filmtransistor (TFT) indirect detection flat panel array with an 18 23-cm active area.
Since that time, several digital mammography systems have been approved by the
FDA in the United States and, in 2010, 70% of all clinical mammography machines
in the United States were digital systems.
Full-Field Digital Mammography

There are compelling advantages to digital mammography, the most important of
which is the ability to overcome the exposure latitude limitations of screen-film
detectors and produce better image quality at lower doses. Other reasons include
FIGURE 8-24 A comparison of the

response of a screen-film and digital radiography detector to variations in relative
exposure shows the large differences in
exposure latitude and image response. The
For Processing image (DICOM term for a raw
image that has been corrected for flaws and
detector non-uniformities) is linearly related
to incident exposure; the For Presentation
image is postprocessed by contrast and spatial resolution enhancements for viewing.
Digital > 200:1
Film
Response
264
For Processing
25:1
Long Relative Exposure
For Presentation
increased technologist productivity through rapid acquisition and evaluation of digital images; reduced patient waiting time, particularly for diagnostic studies; and
improved image quality and conspicuity of lesions.
Screen-film detectors for mammography are excellent image capture devices,
and the film serves as both the display device and the storage media, but the major
drawbacks are the limited exposure dynamic range and narrow latitude, which are
necessary to achieve high radiographic contrast (Fig. 8-24). A highly glandular
breast can produce exposure latitude that exceeds 200:1, causing underexposed
film areas corresponding to the glandular tissue, and overexposed film areas corresponding to the skin line and thinner parts of the breast. Digital flat panel and
digital cassette (CR) detectors have exposure dynamic ranges greater than 1,000:1,
which result in images showing little contrast when the entire ranges of pixel values
are displayed. The full exposure data is contained in the the raw, For Processing
image shown in Figure 8-24. With mammography-specific image postprocessing,
high image contrast can be rendered over all exposure levels (e.g., from the high
detector exposure at the skin line to the low detector exposure behind dense glandular areas of the breast), as shown in the processed For Presentation image. Postprocessing image enhancement is a key element in producing an optimized digital
mammogram.
Flat panel detector arrays, which became clinically available in the mid to late
1990s for general radiography (see Chapter 7), are the major detector technologies
used in digital mammography. An active matrix, thin film transistor (TFT) array collects the local signal (electric charge) generated during the x-ray exposure, absorption, and conversion process; stores the charge in a capacitor attached to each
detector element; and actively reads the array immediately afterward to produce the
image (Fig. 8-25). Key components in each detector element include a transistor
(which serves as an on-off switch), a charge collection electrode, and a storage
capacitor. During image acquisition, the transistor, controlled by the gate line, is in
the off position, and the charge collected by the electrode from incident x-rays is
stored in the capacitor. Immediately after the exposure, readout occurs by activating each gate line, one row at a time, which turns on the transistors along each row,
allowing charge transfer via the drain lines along each column to a series of charge
amplifiers and digitizers. The digital image is then stored in computer memory. In
some fast flat panel designs, readout of the whole array is performed in hundreds
of milliseconds, allowing near real-time acquisition of data for applications such as
digital tomosynthesis.
Gate Drivers
265
FIGURE 8-25 The flat panel array is a twodimensional matrix of individual detector elements lithographed on an amorphous silicon
substrate. Each detector element is comprised
of an electronic switch (the TFT), a charge collection electrode, and a storage capacitor. The
gate and drain lines provide the mechanism to
extract the locally stored charge by activating
gate lines row by row, and collecting charge
down each column. Charge amplifier and
digitizers convert the signals to corresponding
digital values for transfer to the digital image
matrix.
aSilicon
TFT array
Gate lines
IV
TE
RS
DR
Drain lines
GA
TFT
Capacitor
MU
LTIP
LEX
Electrode
ER
Charge amplifier
Common to all flat panel TFT arrays is the amorphous silicon circuit layer
(Fig.8-26A). At present, approved digital mammography systems are based on four
technologies. The first is an indirect x-ray conversion TFT flat panel array receptor with
100-mm sampling pitch and 18 23-cm or 24 31-cm FOV (Fig. 8-26B). Astructured cesium iodide (CsI) phosphor converts x-rays into light; the light is emitted onto
a photodiode in each detector element, from which the charge is generated and stored
in the local capacitor. Indirect conversion describes absorption of x-rays in the CsI, the
production of secondary light photons directed to a photodiode, and the generation of
the charge, which is stored on the storage capacitor in that detector element.
A second technology is based on a direct x-ray conversion TFT detector with 70-mm
sampling pitch and 24 29-cm FOV (Fig. 8-26C). A semiconductor x-ray converter,
Se, is uniformly deposited between two electrodes, with a large voltage placed across
the Se layer. Incident x-rays absorbed by the Se directly produce electron-hole pairs;
the applied voltage causes the electrons to travel to one electrode and the holes to the
opposite electrode. The local storage capacitor captures the charge from the collection
electrode. Direct conversion refers to the direct generation of charge by x-rays within
the photoconductor and capture by the electrode without intermediate signals.
A third technology approved for digital mammography is a cassette-based, dualside readout CR photostimulable storage phosphor (PSP) imaging plate detector
and reader system. Cassettes of 18 24 cm and 24 30 cm are used in place of
screen-film cassettes (Fig. 8-27), whereby x-rays absorbed in the phosphor raise
electrons to higher energy levels in the PSP material, where a fraction of them are
caught in semi-stable electron traps. An exposed imaging plate is subsequently processed by a reader system that scans the plate with a laser beam with an effective
spot size of 50microns. The laser beam provides energy to free the stored excited
electrons from the traps; these electrons fall to a lower energy state with the emission of light, a process called stimulated luminescence. Some of the light reaches
a photomultiplier tube (PMT), which produces and amplifies an electrical current
266

A
Amorphous silicon TFT flatpanel

Charge electrode
ThinFilmTransistor
Storage capacitor
Glass substrate
Indirect detection
Xray
Light
Structured
phosphor (Csl)
Photodiode
Light Charge
Direct detection
Xray
electronhole
High Voltage
pairs
Amorphous
selenium (aSe)
FIGURE 8-26 A. TFT flat panel arrays have a common underlying amorphous silicon structure, but there are
two types of TFT detectors, determined by the conversion of the x-rays into electrical signals. B. Indirect x-ray
conversion TFT arrays have an additional photodiode layer placed on the charge collection electrode of each
detector element. The photodiodes are optically coupled to a layer of a phosphor, such as CsI, that produces
light when irradiated by x-rays. The light produced by x-ray absorption in the phosphor in turn creates mobile
electric charge in the photodiode. The charge is collected by the electrode and stored in the capacitor. C. A
Direct x-ray conversion TFT array has a semiconductor layer of approximately 0.5 mm thickness between
the surface electrode extending over the detector area and each detector element electrode, under a potential
difference of approximately 10 V/mm. As hole-electron pairs are created in the semiconductor layer by x-rays,
the electrons and holes migrate to the positive and negative electrodes, with minimal lateral spread. A proportional charge is stored on the capacitor. The outward appearances of the indirect and direct detectors are
similar, shown by the pictures on the right.
FIGURE 8-27 The CR digital mammography detector uses a cassette that contains a PSP imaging plate,
compatible with screen-film cassette sizes. Workflow is similar to screen film, but the AEC settings require
independent calibration. When exposed to x-rays, the storage phosphor contains a latent image of trapped,
excited electrons. Subsequent imaging plate processing uses dual-side readout (Fig. 8-27, right) by a scanning 0.050-mm laser beam, plate translation, and approximately 60-s readout time. Stimulated luminescence
produces a light signal at each point on the imaging plate that is detected and digitized to create the digital
image. Erasure of the imaging plate after the readout eliminates residual signals prior to next use.
267
proportional to the light intensity. The signal from the PMT is digitized to provide
pixel values of the digital image. Typical turnaround time of an imaging plate is on
the order of 60 to 90 s, similar to the batch mode processing of screen-film images.
The fourth technology approved for digital mammography uses a narrow 22 cm
long charge-coupled-device (CCD)-detector array, optically coupled to a CsI phosphor
x-ray converter, to capture x-ray signals in a highly collimated x-ray beam configuration (a slot). The moving collimator slot sweeps a narrow x-ray beam across the
breast. The detector moves across the image plane to stay aligned with the scanningslot x-ray beam. The scan acquires data in the medial to lateral directions in about 5s,
with 50-mm sampling over a 22 29-cm FOV. A major advantage of the scanningslot fan-beam system is that very little scattered radiation from the breast reaches the
detector. This technology is not being manufactured at the present time.
All current full-field digital mammography detectors have similar attributes with
respect to exposure latitude. In terms of spatial resolution, the direct conversion TFT
detector achieves the best intrinsic spatial resolution due to active charge collection and
minimal signal spread in the semiconductor material, the signal being acquired onto
each 70 mm 70 mm detector element (dexel). The indirect conversion TFT detector
uses a CsI(Tl)-structured phosphor to reduce lateral spread of the light photons onto a
100mm 100mm dexel, and the CR system uses a 50-mm scanning laser beam and
plate translation speed to maintain an effective 50 mm 50 mm dexel. In terms of
detective quantum efficiency (DQEdefined in Chapter 4), the indirect TFT array has
better performance at low to intermediate spatial frequencies, and the direct TFT array
has better performance at higher spatial frequencies. For cassette-based CR mammography detectors, the image quality and SNR at a given incident radiation exposure to the
detector is lower than that of the TFT array detectors. In general, mammography technique factors and the corresponding average glandular doses are lowest for the indirect
TFT detectors and highest for the cassette-based CR detectors. In most situations (with
appropriate calibration of the AEC), use of digital detector systems will result in lower
doses than screen-film detectors, particularly for thick, dense breast tissues.
Due to high-spatial resolution requirements, digital mammography images are
quite large. A digital mammography system with a 0.07-mm pixel, 2 bytes per pixel,
and a 24 29-cm active detector area, produces a 27-MB image. One screening study
(4 images) will therefore contain 108 MB of data and, if 3 years of prior images are
reviewed, 432 MB of data will be required per patient reading session. The network
overhead and storage required for the picture archiving and communications system
are very large and even larger with the increased number of images in a diagnostic
exam. In addition, if For Processing images are stored, another factor of two increase is
incurred. Typical digital image sizes for mammography systems are listed in Table 8-6.
TABLE 8-6 DIGITAL IMAGE SIZE FOR MAMMOGRAPHY SCREENING EXAMS

DETECTOR
TYPE
FOV (cm)
PIXEL
SIZE (mm)
IMAGE
SIZE (MB)
EXAM
SIZE (MB)
3 Y PRIORS (MB)
Indirect TFT
19 23
0.10
35
140
Indirect TFT
24 31
0.10
15
60
240
Direct TFT
18 24
0.07
18
70
280
Direct TFT
24 29
0.07
27
108
432
CR
18 24
0.05
32
128
512
CR
24 30
0.05
50
200
800
268
Currently, the recommendation regarding mammography image compression to

reduce file size is to not use lossy compression algorithmsthat is, an exact (lossless)
image must be reproduced during the decompression step. For most digital mammograms, there is a significant background area that is pixel-padded with a constant
value number in the For Presentation image. On average, a compression factor of 3
to 5 can be achieved using lossless compression, depending on the fraction of the
image matrix area comprising the breast. FDA regulations require that facilities maintain mammography films, digital images, and reports in a permanent medical record
of the patient for a period of not less than 5 years and not less than 10 years if
no additional mammograms of the patient are performed at the facility. A longer
retention period may be mandated by state or local law. Many considerations are
described in the ACR practice guideline, Determinants of Image Quality in Digital
Mammography available on the ACR website, http://www.acr.org.
Processing and Viewing Digital Mammograms

Even the best digital detectors have malfunctioning detector elements, column and
line defects, and variation in gain across the detector area. Without corrections, these
artifacts would result in unacceptable image quality and interfere with the diagnostic
interpretation. Manufacturers use preprocessing algorithms to correct these defects.
First, the locations of inoperative detector elements and line defects are determined,
and these data are interpolated from nearest-neighbor pixels. Next, a uniform beam
of x-rays exposes the detector and generates an image, showing the variable gain from
the detector submodule electronics. Multiple images are averaged together to create a
flat-field correction image, which is normalized and inverted as an inverse-gain map.
When applied to an uncorrected image on a pixel-by-pixel basis, the gain variations are
canceled, and the corrected raw image is defect-free (see Chapter 4). In practice, flatfield gain maps for various acquisition conditions are measured, stored on the system,
and applied to each clinical image that is acquired under similar conditions.
The output For Processing image has wide latitude, digital signals that are proportional to the detector exposure or the logarithm of exposure, and very low contrast
(Fig. 8-28A). While not useful for diagnosis by the radiologist, this is the preferred
image for Computer-Assisted Detection (CADe) programs, which use computer algorithms to identify image features that may indicate the presence of breast cancer. Contrast enhancement of the For Processing image is shown in Figure 8-28B, which looks
similar to a screen-film image. Nonlinear postprocessing algorithms applied to the For
Processing image data identify the skin line of the mammogram, modify the local-area
pixel values, and equalize the response in the high-signal, low-attenuation skin area
to the other areas of the breast. Contrast and spatial resolution enhancements are then
applied to create the For Presentation image for viewing and diagnosis. Unlike film
radiographic contrast, which is variable in the areas corresponding to the thinnest and
thickest parts of the breast image (see Fig. 8-22), image contrast is equalized across the
entire the digital image (Fig.8-28C). Each manufacturer has proprietary methods of
image processing; the radiologist must to adopt a particular look for viewing current
and prior digital mammograms for consistency. The For Processing images should be
saved in addition to the For Presentation images as part of a patients examination if CAD
algorithms will be used in possible subsequent applications, or if other image processing algorithms are to be applied, since For Presentation images cannot be unprocessed.
The d
isadvantage to saving the For Processing images is the extra storage required.
It is strongly recommended that diagnostic interpretation of digital mammograms
be performed on display monitors approved by the FDA for mammography having a
minimum of 5 million pixels (5 MP) and a calibrated, sustained maximal luminance
Detector corrections,
gain map corrections,
no enhancement,
For Processing
Simple linear contrast &

brightness corrections,
no advanced processing
269
FIGURE 8-28 A. The For Processing

images of the right and left cranial-caudal
breast projection images have detector
and gain map corrections applied with
unity radiographic contrast. B. Linear
contrast and brightness adjustments are
applied. Appearance is similar to a screenfilm image. C. Nonlinear processing identifies the breast skin line and densest parts
of the image, to equalize the response of
the processed image, with contrast and
spatial resolution enhancement.
Skin equalization
processing and nonlinear enhancement
For Presentation
of at least 450 cd/m2 (per ACR guidelines for digital mammography), but preferably at
600 cd/m2. Monitors of lower luminance make reading conditions more susceptible to
poor ambient lighting and illuminance conditions. A typical workstation has two largeformat portrait-oriented LCD monitors (each 54 cm diagonal, 34 cm width, 42 cm
height) with 2,560 2,048 pixels at a pixel pitch of approximately 0.165 mm and a
contrast ratio (ratio of brightest to darkest luminance) greater than 350:1. The optimal
viewing distance for the resolution provided by the 0.165-mm pixel is about 50 cm
(slightly less than arms length). A lower resolution, smaller format navigation monitor is used for patient lists and operating system access. All monitors should be calibrated to conform to the DICOM Grayscale Standard Display Function, as described in
Chapter 5, to ensure image contrast is consistently and optimally displayed.
To optimize the workflow of the radiologist, mammography workstations should
provide automated mammography-specific hanging protocols that present the images
in a useful and sequential way for review and diagnosis according to view, laterality,
magnification, and prior examination comparisons. Demographic overlays (patient
name, technique factors, and details of the exam) should be mapped in areas outside
of the breast anatomy. It is important to display images in true size at one point in
the diagnostic review (as close as possible to a magnification of 1, independent of the
detector pixel dimension so that measurements can be made directly on the monitor
if necessary) and to effectively handle images of different size from different manufacturers to appear the same size on the workstation when direct comparisons are
made. Also, viewing of all mammograms at one image pixel to one monitor pixel
ensures that the monitor does not limit the intrinsic resolution of the image. This
requires a zoom and pan procedure for each image. Sometimes, radiologists will
use a magnifying glass on a displayed digital mammogram to overcome limitations of
the human visual system acuity; this is reasonable as long as the image (or portion of)
is displayed at full resolution. However, a digital magnification by pixel replication
and interpolation to enlarge the image directly on the monitor is preferable. Issues
270
confronting digital mammography workflow and hanging protocols are addressed in

the Integrating the Healthcare Enterprise Mammography Image Integration Profile*.
Even after solving many of the display concerns, image appearance may be
considerably different between the manufacturers proprietary processing algorithms,
which can make comparisons challenging.
An alternative to soft-copy reading of digital mammograms is the use of an
FDA-approved laser printer to record all digital studies on film, an unlikely scenario. However, there is FDA requirement to have the capability of printing film
images from digital mammograms at each facility to be able to provide hard-copy
images topatients upon request. QC procedures must be performed on the printer
periodically, as described in the printer manufacturers QC manual for digital
mammography printers.
Computer-Aided Detection and Computer-Assisted

Diagnosis Systems
A computer-aided detection (CADe) system is a computer-based set of algorithms that
incorporates pattern recognition and uses sophisticated matching and similarity rules
to flag possible findings on a digital mammographic image, which may be a digitized
film mammogram. The computer software searches for abnormal areas of density, mass,
calcifications, and/or structural patterns (e.g., spiculated masses, architectural distortion) that may indicate the presence of cancer. The CADe system then marks these
areas on the images, alerting the radiologist for further analysis. For digitally acquired
mammograms, the CADe algorithms require the use of linear, For Processing images
for analysis, as the algorithms are trained on these images and evaluation of processed
images would result in lower performance in terms of sensitivity and specificity.
Computer-assisted diagnosis (CADx) devices include software that provides
information beyond identifying suspicious findings; this additional information
includes an assessment of identified features in terms of the likelihood of the presence or absence of disease, or disease type. For instance, a CADx system for mammography will not only identify clusters of microcalcifications and masses on digital
mammograms but also provide a probability score of the disease or malignancy to the
radiologist for each potential lesion. Currently, CADe and CADx systems are widely
used by radiologists who interpret mammography studies in addition to other duties.
The fraction of False-positive marks (marks placed on image features that do not
correspond to malignancy) can be high and is one of the downsides of using these
systems. Many expert mammography radiologists avoid using CAD for this reason.
Stereotactic Breast Biopsy

Stereotactic breast biopsy systems provide the capability to localize breast lesions
in three dimensions and physically sample them with a targeted biopsy instrument.
Generally, these systems are used for targeting microcalcifications associated with
lesions, whereas localization and biopsy of masses and areas of architectural distortion are performed either with ultrasound guidance or MRI. Figure 8-29A shows
a prone-geometry stereotactic imaging system. The image receptors of these systems are CCD cameras coupled to x-ray phosphor screens of 5 cm 5-cm FOV by
either lens optics or fiber optics. A 1,000 1,000 detector element matrix, with an
*http://www.ihe.net/Resources/handbook.cfm)
Caudal
271
Cranial
Detector
Breast
Table
X-ray tube
C +15 image
0 image
15 image
Compression
x-shift
Caudal
2
1 Image detector
Breast support
Detector
to lesion
distance
1
2
1
2
Biopsy gun
location
Composite
Lesion shift = opposite side/2
1
2
=15
Compression paddle
Cranial
Posterior-Anterior
perspective
X-ray tube shift
15
+15
0
tanq =
opposite
shift / 2
=
adjacent Zdetector to lesion
Detector to lesion
shift(mm)
distance = adjacent side
2 tan 15
= 1.866 shift (mm) of right triangle
Zdetector to lesion =
FIGURE 8-29 A. A stereotactic biopsy system with a prone positioning system is shown. The patients
pendulant breast is positioned on the detector platform and compressed by an open-access paddle. The x-ray
tube pivots about a fulcrum point to acquire digital images at 15-, and 15-degree projections within the
5 5-cm detector FOV. B. Breast lesions positioned more cranially (toward the tube) will have a greater shift
in the image pair than lesions closer to the detector. C. Trigonometric relationships allow the determination
of lesion depth from the measured shift distance in the images. A frame of reference calibration between the
images on the computer and the biopsy gun placed on rails between the x-ray tube and breast allows a trajectory calculation and needle orientation under computer control.
e ffective detector element size of 50 mm (0.05 mm), which is equivalent to10 lp/mm,
is deemed sufficient to visualize microcalcifications. Image acquisition is performed
by integrating light from x-ray exposure of the phosphor screen on the CCD photodetector, reading the detector element array one line at a time after the exposure,
amplifying the e lectronic signal, and digitizing the output.
The breast is aligned and compressed by an open-area paddle. After verification
with a scout image, two images of the breast are acquired, at 15 and 15 degrees
from the normal, causing the lesion projections to shift on the detector, as shown in
Figure 8-29B. The shift distance is measured (Fig. 8-29C) from trigonometry relationships, the opposite sides of two right triangles, and the common adjacent side,
each with a 15-degree angle (30 degrees total). The adjacent side represents the
lesion distance from the detector, and since Opposite/Adjacent = tan( = 15 degrees),
the lesion distance from the detector (mm) is calculated as 1.866 shift (mm). Trajectory information is determined by the position of the marked points in the images
correlated with the frame of reference of the biopsy device, which is positioned and
activated to capture the tissue samples.
Digital add-on biopsy units with larger FOVs (e.g., 8 8 cm) using CCDs or
other photosensitive detectors that fit into the cassette assemblies of the standard
mammography units are alternatives to the prone biopsy table, as well as full-field
digital mammography units that can provide similar capabilities.
272
Breast Digital Tomosynthesis

Digital radiographic imaging offers postacquisition processing capabilities that are
not possible with conventional analog imaging systems. One of the major problems with conventional projection imaging is that overlying and underlying anatomy are superimposed on the pathology, often obscuring the visualization and
detection of the cancer or other abnormality. A method for reducing this superimposition is to acquire multiple low-dose images at several angular positions as
the x-ray tube moves in an arc about the breast. Each image projects the content
in the breast volume with different shifts depending on the distance of the object
from the detector. With high-speed digital readout, several projection images can
be acquired in under 10 s, and each image set is processed with a limited-angle
reconstruction algorithm to synthesize a tomogram (in-focus plane) at a given
depth in the breast. Many tomograms representing incremental depths within
the volume are produced, and in-focus planes throughout the breast can assist
in enhanced detection of pathology (e.g., tumor) or elimination of superimposed
anatomy that mimics pathology. Even though many more images must be reviewed
by the radiologist, digital tomosynthesis technology can lead to a superior diagnosis that might save a patient an unneeded biopsy or provide guidance for early
treatment of a cancer.
Digital tomosynthesis for breast imaging is made possible with fast readout TFT
arrays by acquiring many sequential images (11 to 51) over a limited angle ( 7.5
to 50 degrees) in a short time span (4 to 20 s). For one vendors implementation,
15 projection images of the compressed breast are acquired over a range of
15degrees (1 image per degree) in 4s (270 ms per image), using a W target with
0.7-mm Al filtered x-ray beam, 32 to 38 kV, with no antiscatter grid, and 2 2
pixel binning of the digital detector array (140-mm sampling). The image acquisition process is illustrated in Figure 8-30A. Average glandular dose (discussed
below) for 15 image acquisitions is approximately 2 mGy for the accreditation
phantom. Larger breast thicknesses will require larger doses for tomosynthesis
acquisition, but the dose is similar to that of a single view mammogram of the
same compressed thickness. The 15 images are used in the reconstruction process
to create 1-mm focal plane images (tomograms), where the location of the reconstructed plane depends on the reconstruction algorithm, and the number of reconstructed images depends on the compressed breast thickness. Reconstruction can
be performed in several ways: (a) shift and add method shown in Figure 8-30B is
the simplest; (b) filtered backprojection; (c) simultaneous algebraic reconstruction
techniques; and (d) maximum likelihood algorithms. More sophisticated methods
require longer reconstruction times but provide enhancement by further reducing
out-of-plane signals.
A series of approximately 60 tomograms was reconstructed from a volunteer
patient who was imaged in the cranial-caudal and medial-lateral oblique projections; four images from that dataset, at the entrance and exit points of the x-ray
beam and at two depths, are shown in Figure 8.30C. Rapid stack mode viewing
helps the perception of otherwise undetected lesions with good conspicuity. Investigations of breast tomosynthesis have demonstrated that its advantages are greater
for masses and architectural distortion, than for calcifications. The current role of
tomosynthesis is to supplement full-field digital mammography exams; however,
future limited-angle acquisition techniques and processing algorithms are expected
to provide both the tomosynthesis images at in-focal planes throughout the breast
and a high-quality reconstructed conventional projection mammogram from the
same image dataset.
A
Image Acquisition:
Multiple angular views
11 25 images, < 5 s
B Image Reconstruction:
Shift images to select plane
Add images to create tomograms
Left
Low dose image (~15)
273
Right
Tomograms showing two planes
Caudal
Cranial
Medial
Lateral
CC
View
MLO
View
C
FIGURE 8-30 A. Digital tomosynthesis images are acquired using low-dose techniques over a limited
projection angle, from 7.5 degrees up to 30 degrees, with 15 to 40 images comprising the dataset.
Illustrated are a square and circle at two depths within the breast B. Reconstruction of tomogram
images at 1-mm incremental planes in the breast is performed by shifting the acquired images a known
amount, summing the images by projection and normalizing the output. In-focus images are shown for
two depths, illustrating the ability to blur the underlying and overlying signals for the white rectangle
and black circle. C. Tomosynthesis images from a volunteer patient are shown for the cranial-caudal and
medial-lateral oblique projections at 4 depths. The overall breast dose is similar to that from a single
projection mammogram.
274
8.6
Radiation Dosimetry
x-ray mammography is the technique of choice for detecting nonpalpable breast
cancers. However, the risk of carcinogenesis from the radiation dose to the breast
is of concern, particularly in screening, because of the very large number of women
receiving the exams. Thus, the optimization of breast dose is important and dose
monitoring is required yearly as specified by the MQSA regulations.
The glandular tissue is the site of carcinogenesis, and thus the preferred dose
index is the average glandular dose. Because the glandular tissues receive varying
doses depending on their depths from the skin entrance site of the x-ray beam, estimating the dose is not trivial. The midbreast dose, the dose delivered to the plane of
tissue in the middle of the breast, was the radiation dosimetry benchmark until the
late 1970s. The midbreast dose is typically lower than the average glandular dose and
does not account for variation in breast tissue composition.
Average Glandular Dose

The average glandular dose, Dg, is calculated from the equation
Dg = DgN XESAK
where XESAK is the entrance skin air kerma (ESAK) in mGy, and DgN is an air kerma
to average glandular dose conversion factor with units of mGy dose/mGy incident
air kerma. An air ionization chamber is used to measure the air kerma for a given
kV, mAs, and beam quality. Previously, this calculation involved a roentgen to rad
conversion, where the units of the DgN were mrad/R.
The conversion factor DgN is determined by computer simulation methods
and depends on radiation quality (kV and HVL), x-ray tube target material, filter
material, breast thickness, and tissue composition. Table 8-7 lists DgN values for a
50% adipose, 50% glandular breast that is 4.5-cm thick. The coefficients are shown as
a function of HVL and kV for a Mo/Mo target-filter combination. For a 26-kV technique with a HVL of 0.35-mm Al, the average glandular dose is approximately 19%
of the measured ESAK (Table 8-7). For higher average x-ray energies (e.g., Mo/Rh,
Rh/Rh, W/Rh, W/Ag), conversion tables specific to the generated x-ray spectrum must
be used, as the DgN values increase due to the higher effective energy of the beam.
DgN values decrease with an increase in breast thickness for constant beam quality
and breast composition. This is because the glandular tissues furthest from the beam
entrance receive much less dose in the thicker breast due to attenuation (e.g., DgN =
0.220 mGy dose/mGy air kerma for 3-cm thickness versus 0.110 mGy dose/mGy air
Kerma for 6-cm thickness). However, the lower DgN coefficients for the thicker breast
do not mean that larger breasts receive less dose. The lower DgN values are more
than offset by a large increase in the entrance kerma necessary to achieve the desired
OD on the film or SNR in a digital image. Measured average glandular doses for
50% glandular/50% adipose tissue of 2 to 8-cm tissue thicknesses using AEC are
shown in Figure 8-31 for screen-film and direct-detection digital systems.
Factors Affecting Breast Dose

Many variables affect breast dose. The speed of the screen-film receptor, the film
OD, and the digital detector SNR level are major factors. Breast thickness and
tissue composition strongly affect x-ray absorption. Higher kV (higher HVL)
275
TABLE 8-7DgN CONVERSION FACTOR (mGy AVERAGE GLANDULAR DOSE
PER mGy INCIDENT AIR KERMA) AS A FUNCTION OF HVL AND

kV FOR Mo TARGET AND FILTER, 4.5-cm BREAST T
HICKNESS
OF 50% GLANDULAR AND 50% ADIPOSE BREAST TISSUE
COMPOSITION
kV
HVL (mm)
25
0.25
0.140
26
27
28
29
30
0.26
0.144
0.147
0.27
0.149
0.151
0.153
0.28
0.153
0.156
0.158
0.159
0.29
0.159
0.161
0.163
0.164
0.165
0.30
0.164
0.166
0.167
0.168
0.169
0.171
0.31
0.168
0.171
0.172
0.173
0.174
0.175
31
32
0.176
0.32
0.173
0.175
0.176
0.177
0.179
0.181
0.182
0.183
0.33
0.177
0.180
0.181
0.182
0.183
0.185
0.187
0.188
0.34
0.183
0.184
0.185
0.187
0.188
0.190
0.191
0.192
0.35
0.188
0.190
0.191
0.192
0.194
0.195
0.196
0.197
0.36
0.192
0.37
0.38
0.39
0.40
0.195
0.196
0.197
0.198
0.199
0.200
0.202
0.199
0.200
0.202
0.203
0.204
0.204
0.205
0.205
0.206
0.207
0.208
0.208
0.210
0.211
0.212
0.213
0.213
0.214
0.216
0.218
0.219
0.220
Adapted from ACR QC Manual, 1999; converted from mrad per R using 0.01 mGy/mrad and 0.1145 mGy air
Kerma/R
FIGURE 8-31 Approximate average glandular doses (25 to 35 kV using AEC acquisition) are shown for a screen-film detector
with Mo target and Mo/Rh filtration (green),
a digital system using a-Se direct conversion
detector with Mo target and Mo/Rh filtration (purple), and a digital system using a-Se
direct conversion detector with W target
and Rh/Ag filtration (blue). The accreditation
phantom (4.5 cm thickness) is indicated with
the vertical black bar for comparison of average glandular dose noted in MQSA reports.
276
TABLE 8-8AVERAGE GLANDULAR DOSE IN mGy WITH TWO TISSUE
OMPOSITIONS OF THE BREAST USING kV VALUES IN

C
TABLE 8-4a FOR SCREEN-FILM AND TABLE 8-5b FOR DIGITAL
ACQUISITIONS
BREAST THICKNESS (cm)
DETECTOR: BREAST COMPOSITION
Screen-Film: 100% Adipose
0.50
1.25
3.00
4.30
Screen-Film: 50/50
0.65
1.80
3.80
5.20
Digital Selenium, W target: 100%

Adipose
0.50
0.75
1.55
2.45
Digital Selenium, W target: 50/50
0.60
1.00
2.40
4.20
Mo/Mo target/filter, 5:1 grid, and film OD = 1.80 (Fuji AD-M film and AD-fine screens).
W/Rh (2, 4, 6 cm); W/Ag (8 cm) for selenium detector.
increases beam penetrability (lower ESE and therefore lower average glandular
dose) but also decreases subject contrast. Antiscatter grids improve subject and
radiographic contrast, but increase radiation dose by the Bucky or digital technique compensation factor (~2). Table 8-8 lists measured average glandular
doses for a contemporary clinical screen-film mammography system with a film
OD of 1.8, a Mo/Mo target/filter, and a kV recommended in Table 8-4, compared to a digital mammography system utilizing a W/Rh or W/Ag target/filter
combination with kV recommended in Table 8-5. Digital acquisition with higher
effective energy beams isshown to substantially reduce average glandular dose
over that of s creen-filmsystems.
The MQSA regulations limit the average glandular dose for a compressed breast
thickness of 4.2 cm and a breast composition of 50% glandular and 50% adipose
tissue (the MQSA-approved mammography phantom) to 3 mGy per film or digital
image (6mGy for two films or images). If the average glandular dose exceeds 3 mGy
for this size breast, mammography may not be performed. The average glandular
dose using screen-film receptors is typically about 2 mGy per view or 4 mGy for two
views. For full-field digital mammography systems, the dose is lower, typically from
1.2 to 1.8mGy per view.
Mammography image quality strongly depends on beam energy, quantum mottle, and detected x-ray scatter. Faster screen-film detectors can use a lower dose, but
often at the expense of spatial resolution. Digital detectors with higher quantum
detection efficiency can deliver significantly reduced dose; the tradeoff for reduced
dose can be a loss of SNR and contrast-to-noise ratio (CNR) and potentially less
diagnostic information. Higher energy beams (e.g., Rh target Rh filter or W target Rh
filter or Ag filter) can reduce the dose to thick or dense glandular breasts with little
or no loss of image quality.
8.7
Regulatory Requirements
Regulations mandated by the MQSA of 1992 specify operational and technical requirements necessary to perform mammography in the United States. These regulations
are contained in Title 21 of the Code of Federal Regulations, Part 900.
277
Accreditation and Certification

Accreditation and certification are two separate processes. For a facility to perform
mammography legally under the MQSA, it must be accredited and certified. To begin
the process, the facility must first apply for accreditation from an accreditation body
(the ACR or one of several states if located in one of those states). The accreditation
body verifies that the mammography facility meets standards set forth by the MQSA
to ensure safe, reliable, high-quality, and accurate mammography. These include
the initial qualifications and continuing experience and education of interpreting
physicians, technologists, and physicists involved in the mammography program.
Existence of an active QC program and image quality standards are verified and
validated. Certification is the approval of a facility by the US FDA to provide mammography services and is granted when accreditation is achieved.
Specific Technical Requirements and Quality Control

For screen-film mammography systems, requirements are largely unchanged from
the 1999 FDA regulations. Tables 8-9 and 8-10 summarize the annual QC tests
performed by the technologist and physicist, respectively, as required by MQSA
regulations.
Under FDA rules, facilities using x-ray systems other than screen-film mammography must maintain a QA and QC program that is substantially equivalent to the
program recommended by the manufacturer. This has created a number of unique
tests that the technologist and physicist must understand and perform according to
the manufacturers manuals and procedural methods. In addition, QC procedures
for the diagnostic review workstation monitor and the laser printer for producing
hard-copy film are also under the auspices of the manufacturer.
Federal regulations specify that whenever QC test results fall outside of the action
limits for image receptor modalities other than screen-film, the source of the problem
shall be identified and corrective action taken before any further examinations are
performed. However, an alternative QC standard lists three different categories in
which specific action requirements are detailed.
Category A QC tests and action levels are specific to the digital acquisition system
and, when outside of the prescribed limits, the source of the problem must be identified and corrective action taken before any examinations can be performed. These
tests include system resolution, breast dose, image quality evaluation, SNR/CNR,
flat-field calibration, and compression.
Category B QC tests and action levels are specific to the performance of a diagnostic
device used for mammographic interpretation, including the review workstation and
laser film printer. When tests produce results that fall outside of the action limits
specified by the manufacturer, the source of the problem shall be identified and corrective action taken before that specific device can be used for interpretation. Image
acquisition can continue, as long as there are alternate approved diagnostic devices
available for interpretation.
Category C QC tests and action levels evaluate performance of components other than
the digital image receptor or diagnostic devices for interpretation. When test results
are outside the action limits specified by the manufacturer, the source of the problem
shall be identified and corrective action shall be taken within 30 days of the test
date. Clinical imaging and mammographic image interpretation can continue during
this period. Applicable QC tests include collimation assessment, artifact evaluation,
278
TABLE 8-9PERIODIC TESTS PERFORMED BY THE QC TECHNOLOGIST FOR
SCREEN-FILM DETECTOR SYSTEMS

TEST AND
FREQUENCY
REQUIREMENTS FOR
ACCEPTABLE OPERATION
DOCUMENTATION
GUIDANCE
TIMING OF
CORRECTIVE ACTION
Daily: Processor
Established operating
level
0.03 OD
0.15 OD
0.15 OD
QC records 12 mo
QC films 30 d
Before any clinical

film processing is
performed
QC records 12 mo
Phantom images
12wk
Before any
further exams are
performed
QC records since last

inspection
30 d

inspection
inspection, three
previous contact tests
inspection or past
three tests
Before any clinical

film processing
Before any further
exams performed
using cassettes
Before exams
performed using
compression device
Base fog density

Middensity
Density difference
Weekly: Phantom
Center OD
Reproducibility
Density difference
between disk and
background
Phantom scoring
Established operating
level
OD 1.20
0.20 OD
0.05 OD
Four fibers, three speck
groups, three masses
Quarterly:
Fixer retention;
quarterly
Repeat analysis;
quarterly
Residual fixer no
>5mg/cm2
<2% change up or
down from previous
Semiannually:
Darkroom fog
OD difference 0.05
Screen-film
contact
Compression
device
40 mesh copper screen

with no obvious
artifacts
Compression force
111newtons (25 lb)
kV accuracy, HVL assessment, AEC performance and reproducibility, radiation output rate, and compression thickness indicator.
An effort to standardize QC in digital mammography is underway, under the
auspices of the ACR, based upon review of QC test results and findings from the
multi-institutional Digital Mammography Imaging Screening Trial completed in 2005.
The Mammography Accreditation Phantom

The mammography accreditation phantom simulates the radiographic attenuation of
an average-size compressed breast and contains structures that model very basic image
characteristics of breast parenchyma and cancer. Its role is to help determine the adequacy of the overall imaging system (including film processing and digital processing)
in terms of detection of subtle radiographic findings, and to assess the reproducibly of
image characteristics (e.g., contrast and OD or SNR and CNR) over time. It is composed
of a poly-methyl-methacrylate (PMMA) block, a wax insert, and a PMMA disk (4-mm
thick, 10-mm diameter) attached to the top of the phantom, and is intended to mimic
the attenuation characteristics of a standard breast of 4.2-cm compressed breast thickness of 50% adipose and 50% glandular tissue composition. The wax insert contains six
279
TABLE 8-10SUMMARY TABLE OF ANNUAL QC TESTS BY THE MEDICAL
PHYSICIST FOR SCREEN-FILM DETECTOR SYSTEMS

TIMING OF
CORRECTIVE ACTION
TEST
REGULATORY ACTION LEVELS
AEC
OD exceeds the mean by >0.30

(over 26 cm range) or the phantom
image density at the center is <1.20
30 d
kV
>5% of indicated or selected kV;

C.O.V. > 0.02
30 d
Focal spot
See Table 8.1
30 d
HVL
See Tables 8.2 and 8.3
30 d
Air kerma and AEC

reproducibility
Reproducibility C.O.V.a > 0.05
30 d
Dose
>3.0 mGy per exposure
Immediate, before
any further exams
X-ray field/light field

congruence
>2% SID at chest wall
30 d
Compression device
alignment
Paddle visible on image
30 d
Screen speed
uniformity
O.D. variation >0.30 from the

maximum to minimum
30 d
System artifacts
Determined by physicist
30 d
Radiation output
<6.1 mGy/s (700 mR/s)
30 d
Automatic
decompression control
Failure of override or manual release
30 d
Any applicable annual

new modality tests
To be determined
Immediate, before
any further exams
Note: documentation requires the inclusion of two most recent reports

a
C.O.V., coefficient of variation, equal to the standard deviation of a series of measurements divided by the
mean.
AEC, automatic exposure control; SID, source-to-image distance.
cylindrical nylon fibers of decreasing diameter, five simulated calcification groups with
simulated calcifications (Al2O3 specks) of decreasing size, and five low contrast disks of
decreasing diameter and thickness that simulate masses (Fig.8-32A). Identification of
the smallest objects of each type that are visible in the phantom image indicates system
performance. To pass the MQSA image quality standards for screen-film mammography, at least four fibers, three calcification groups, and three masses must be clearly visible (with no obvious artifacts) at an average glandular dose of less than 3 mGy. Further
details on other parameters can be found in the MQSA regulations.
For digital mammography, each manufacturer has set the criteria for meeting an
acceptable performance level in terms of qualitative visibility of the objects in the
accreditation phantom, as well as specific tests using the quantitative evaluation of
regions of interest (ROI) to calculate SNR and CNR. The direct flat panel detector
manufacturer, for instance, requires the visibility of five fibers, four speck groups, and
four masses, whereas the indirect flat panel and the cassette-based CR detector manufacturers require the visibility of four fibers, three speck groups, and three masses.
Quantitative evaluation for one manufacturer includes the evaluation of the signal
under the added acrylic disk and in the background area using ROIs of a specified size
280
FIGURE 8-32 A. The mammography accreditation phantom contains a wax insert which has six nylon fibers,
five aluminum oxide speck groups (six specks in each group), and five disks simulating masses. Mean diameters
of the objects are indicated. The corresponding radiograph of the wax insert shows their undegraded radiographic appearance. B. The wax insert is placed into a cutout in a clear acrylic phantom with dimensions listed in
the diagram. The phantom is intended to mimic the attenuation characteristics of a standard breast of 4.2-cm
compressed breast thickness of 50% adipose and 50% glandular tissue composition. Note the contrast disk
on top of the phantom, which generates a signal for evaluation of contrast. C.Mammographic image of the
composite phantom with scatter and noise illustrates the loss of image quality.
281
and locations, while another uses the largest mass in the ACR phantom and an adjacent
background area. The SNR is calculated as the average signal in the background ROI
divided by its standard deviation. The CNR is determined as the difference of the average signals from the object and background, divided by the standard deviation in the
background ROI. Each manufacturer requires the mammography site technologist to
measure the SNR and CNR weekly using the accreditation phantom, and to verify that
the calculated values are within the manufacturers established limits. When found out
of tolerance, the system may not be used clinically until the causes of the failures are
determined and the corrections are implemented and verified with subsequent tests.
Quality Assurance
Mammography is a very demanding imaging procedure, in terms of equipment performance, and in regards to technical competence of the radiologist, technologist,
and medical physicist. Achieving the full potential of the mammography examination
requires careful optimization of the technique and equipment. Even a small change
in equipment performance, film processing, digital image SNR or CNR, image artifacts, patient setup, or image-viewing conditions can decrease the sensitivity and
specificity of mammography.
Ultimate responsibility for mammography quality assurance rests with the radiologist in charge of the mammography practice, who must ensure that all interpreting
radiologists, mammography technologists, and the medical physicist meet the initial
qualifications and maintain the continuing education and experience required by the
MQSA regulations. Records must be maintained of employee qualifications, quality
assurance, and the physicists tests. Clinical performance and QC issues should be
periodically reviewed with the entire mammography staff.
The technologist performs the examinations, that is, patient positioning, compression, image acquisition, and film processing and/or evaluation of digital image
quality. The technologist also performs the daily, weekly, monthly, and semiannual
QC tests. The medical physicist is responsible for equipment performance measurements before first clinical use, annually thereafter, and following major repairs, and
for oversight of the QC program performed by the technologist.
SUGGESTED READING
ACR Mammography Quality Control Manual, 1999. American College of Radiology, Reston, VA.
ACR Practice Guideline for the Performance of Screening and Diagnostic Mammography. 2008.
American College of Radiology, Reston, VA. Document is available at: http://www.acr.org/
SecondaryMainMenuCategories/quality_safety/guidelines/breast/Screening_Diagnostic.aspx
Food and Drug Administration Mammography Program (MQSA). The MQSA regulations and other
documentation are available on the Web at http://www.fda.gov/cdrh/mammography
Integrating the Healthcare Enterprise (IHE) Radiology Technical Framework. Mammography image
(MAMMO) profile. Document is available at: http://www.ihe.net/Technical_Framework/upload/IHE_
Rad1344TF_Suppl_MAMMO_T1_2006-04-13.pdf
Pisano E, Gatsonis C, Hendrick E, et.al. Performance of digital versus film mammography for breast-cancer
screening. N Engl J Med 2005;353:17731783.
Practice Guideline for Determinants of Image Quality in Digital Mammography. 2007. American College of
Radiology, Reston, VA. Document is available at: http://www.acr.org/SecondaryMainMenuCategories/
quality_safety/guidelines/breast/image_quality_digital_mammo.aspx
Chapter
Fluoroscopy
Fluoroscopy systems produce projection x-ray images and allow real-time x-ray
viewing of the patient with high temporal resolution. They are commonly used to
position the imaging system for the recording of images (e.g., angiography) and to
provide imaging guidance for interventional procedures (e.g., angioplasty). Before
the late 1950s, fluoroscopy was performed in a darkened room with the radiologist
viewing the faint scintillations from a thick fluorescent screen. Modern fluoroscopic
systems use image intensifiers (IIs) coupled to digital video systems or flat panel
digital detectors as image receptors. Fluoroscopy has undergone much technological
advancement in recent years. The II has increased in size from the early 15-cm
(6-inch)-diameter field of view (FOV) systems to 40-cm (16-inch) systems available
today. Analog television (TV) cameras have been replaced with high-resolution, lownoise, digital charge-coupled device (CCD) or complementary metal-oxide semiconductor (CMOS) cameras. Flat panel detector technology has led to larger rectangular
FOV systems (e.g. 48 cm) with high spatial resolution and improved image fidelity,
and will gradually replace II detectors. Improvements in fluoroscopic systems with
increased x-ray beam filtration, higher x-ray detection efficiency phosphors, lower
acquisition frame rates, and electronic image guidance and enhancement have led to
lower dose operation. Now, dose-monitoring technology is a component of all interventional fluoroscopy systems manufactured since June 10, 2006.
9.1 Functionality
Real-time imaging is usually considered to be 30 frames per second (FPS), equivalent to older analog television frame rates in the United States, and sufficient to
provide the appearance of continuous motion. Modern general-purpose fluoroscopy
systems use a pulsed x-ray beam in conjunction with digital image acquisition, which
commonly allows variable frame rates ranging from 3 to 30 FPS. Lower frame rates
reduce radiation dose when high temporal resolution is not needed. Digital fluoroscopy systems allow the recording of a real-time sequence of digital images (digital
video) that can be played back as a movie loop with subsequent archiving in the
patients electronic file in the picture archiving and communications system (PACS).
Unrecorded fluoroscopy sequences are used for advancing a catheter (positioning)
during angiographic procedures and, once the catheter is in place, an image sequence
is recorded using high frame rate pulsed fluorography as x-ray contrast material is
injected through the catheter into the vessels (e.g., angiography) or the body cavity of
interest. In gastrointestinal fluoroscopy, much of the diagnosis is formulated during
the unrecorded fluoroscopy sequence; however, fluorographic images are acquired
for documenting the study and illustrating important diagnostic findings. Imaging
the heart at rates of 15 to 60 FPS, formerly obtained with cine cameras on the optical
distributor of the II, using 35 mm continuous roll photographic film is now achieved
by using digital cameras or directly with flat-panel detectors.
282
Chapter 9 Fluoroscopy
283
9.2 Fluoroscopic Imaging Chain Components

A standard fluoroscopic imaging chain is shown in Figure 9-1. The added filtration
in fluoroscopy systems has increased in recent years, as pulsed systems with high
mA capabilities have replaced older continuous fluoroscopy systems. Copper filtration combined with lower kV values allows angiography systems to provide lower
dose operation while still delivering high image contrast for angiographic applications. Collimators on fluoroscopy systems utilize circular diaphragms for II-based
systems and rectangular collimation for flat panel fluoroscopy systems. The collimators on fluoroscopy systems are motorized, which allows the collimated x-ray beam
to change size in response to adjustment of the source-to-image distance (SID) or
when the FOV is electronically changed (see magnification mode in Section 9.3).
In addition to added filtration and beam collimation, some fluoroscopy systems
also have operator adjustable attenuating wedges to provide additional attenuation at
specified locations in the x-ray field, such as the pulmonary space between the heart
and chest wall, where excessive x-ray penetration of the pulmonary structures can
lead to bright regions in the image. The use of wedges can reduce the glare from these
areas, equalize the x-ray beam incident on the image receptor, and lower radiation
dose to the patient.
The principal feature of the imaging chain that distinguishes fluoroscopy from
radiography is the ability to produce real-time x-ray images with high frame rates
and a low-dose per image. In general, five configurations of fluoroscopy systems are
available: gastrointestinal/genitourinary (GI/GU) systems (with the x-ray tube under
the patient table); remote fluoroscopy (over-table x-ray tube); interventional vascular
(vascular surgery, radiology); cardiology and interventional electrophysiology, and
mobile C-arm configurations (surgery, pain medicine). All configurations have the
same basic components.
The basic product of a fluoroscopic imaging system is a projection x-ray image;
however, a 10-minute on time fluoroscopic procedure, if conducted at 30 FPS,
produces a total of 18,000 individual images! Due to the extremely large number
of images required to depict motion and to achieve positioning, fluoroscopic systems must produce each usable image with about one thousandth of the x-ray dose
of radiography, for radiation dose reasons. Consequently, a very sensitive low-noise
Camera Electronics
CCD or CMOS chip
Optical Coupling
ABC Sensor
lmage lntensifier
Anti-scatter Grid
KAP Meter
Collimator
Filters
X-ray Tube
Figure 9-1 The fluoroscopic imaging chain is illustrated, with the patient in
thesupine position. This figure shows an II as
the detector, but flat panel detector systems
are common as well. The detector in fluoroscopy is more sensitive than radiography to
enable low-dose, real-time, longer duration
studies. The x-ray system includes a collimator with motorized blades to adjust to the
FOV, and a Kerma Air Product (KAP) meter to
monitor radiation output for each patient is
often available on interventional and cardiac
catheterization fluoroscopic systems.
284
etector is needed. IIsand fluoroscopic flat panel detectors operate in a mode that is
d
several thousand times more sensitive than a standard radiographic detector and, in
principle, can produce images using several thousand times less radiation. In practice, standard fluoroscopy uses about 9 to 17 nGy (1 to 2 mR) incident upon
the detector per image, whereas a computed radiography (CR) system requires an
exposure of about 5 to 9 mGy (0.6 to 1 mR) to the detector.
9.3 Fluoroscopic Detector Systems
The Image Intensifier, Based Fluoroscopy Systems

A diagram of a modern II is shown in Figure 9-2. There are four principal components of an II: (a) a vacuum housing to keep air out and allow unimpeded electron
flow, (b) an input layer that converts the absorbed incident x-rays into light, which in
turn releases electrons, (c) an electron optics system that accelerates and focuses the
electrons emitted by the input layer onto the output layer, and (d) an output phosphor that converts the accelerated electrons into a visible light image. When appropriate, an antiscatter grid is mounted adjacent to the input layer on the housing.
Input Screen
The input screen of the II consists of four different layers, as shown in Figure 9-3. The
first layer is the vacuum window, a thin (typically 1 mm) aluminum window that is
part of the vacuum containment vessel. The vacuum window keeps the air out of the
II, and its curvature is designed to withstand the force of the air pressing againstit.
The vacuum window of a large FOV (35-cm) II supports over a ton of force from
atmospheric air pressure. A vacuum is necessary in all devices in which electrons are
accelerated across open space. The second layer is the support layer, which supports
the input phosphor and photocathode layers. The support, commonly 0.5mm of
aluminum, is the first component in the electronic lens system, and its curvature is
designed for accurate electron focusing.
Figure 9-2 A diagram of the modern II is shown. X-rays strike the input phosphor, producing light that stimulates the photocathode to emit electrons, which are ejected into the electronic lens system. The electronic lens
system is comprised of the cathode, anode, and three additional focusing electrodes (G1, G2, and G3). Energetic
electrons strike the output phosphor, generating light that is then imaged by a digital camera system.
285
Electrons
Photocathode
CsI Needles
Support
Vacuum Window
X-ray
Light Emission
in CsI Needle
Figure 9-3 A detailed view of the input window of an II is shown. Incident x-rays pass through the
antiscatter grid (not shown), and must also pass through the vacuum support window (,1 mm Al) and the
support for the input phosphor (,0.5 mm Al). X-rays are absorbed by the CsI input phosphor, producing green
light that is channeled within the long CsI needles to the antimony trisulfide photocathode, which then ejects
electrons into the vacuum space of the II. Scanning electron micrographs illustrate the crystalline needlelike
structure of CsI.
After passing through the Al input window and substrate, x-rays strike the input
phosphor, whose function is to absorb the x-rays and convert their energy into visible light. The input phosphor must be thick enough to absorb a large fraction of the
incident x-rays, but thin enough to not significantly degrade the spatial resolution
of the image by the lateral dispersion of light through the phosphor. Virtually all
modern IIs use cesium iodide (CsI) for the input phosphor. With proper deposition techniques, CsI can be made to form in long, columnar crystals (Fig. 9-3). The
long, thin columns of CsI function as light pipes, channeling the visible light emitted
within them toward the photocathode with less lateral spreading than an amorphous
phosphor. As a result, the CsI input phosphor can be quite thick and still produce
high resolution. The CsI crystals are approximately 400 mm tall and 5 mm in diameter and are formed by vacuum deposition of CsI onto the substrate. The CsI crystals
have a trace amount of sodium, causing it to emit blue light. For a 60-keV x-ray photon absorbed in the phosphor, approximately 3,000 light photons (at about 420 nm
wavelength) are emitted. The K-edges of cesium (36 keV) and iodine (33 keV) are
well positioned with respect to the fluoroscopic x-ray spectrum, which contribute to
high x-ray absorption efficiency.
The fourth layer of the input screen is the photocathode, which is a thin layer of
antimony and alkali metals (such as Sb2S3) that emits electrons when struck by visible
light. With 10% to 20% conversion efficiency, approximately 400 electrons are released
from the photocathode for each 60-keV x-ray photon absorbed in the phosphor.
Electron Optics
X-rays are converted into light which then ejects electrons from the input screen into
the evacuated volume of the II. The electrons are accelerated by a large electric field
created by a high voltage- between the anode and the cathode. Focusing the electrons is
accomplished by the several electrodes in the electron optics chain. The kinetic energy
of each electron is dramatically increased by acceleration due to the voltage difference
between the cathode and anode, resulting in electronic gain. The spatial pattern of electrons released at the photocathode (Fig. 9-3) is maintained at the output phosphor,
albeit minified. The curved surface of the input screen, necessary for proper electron
286

Flat Output Surface
Curved lnput Surface
lnput lmage
Output lmage
Figure 9-4 The input phosphor/photocathode surface is curved to accommodate the physics of electronic
focusing, and the output phosphor is planar. Because the image is produced on a curved surface and projected
onto a plane surface, a characteristic pincushion distortion results in the output image.
focusing, causes unavoidable pincushion distortion of the image (Fig. 9-4). TheG1, G2,
and G3 electrodes (Fig. 9-2), along with the input phosphor substrate (the cathode) and
the anode just proximal to the output phosphor, comprise the five-component (pentode) electronic lens system of the II. The electrons are released from the photocathode
with very little kinetic energy, but under the influence of the 25,000 to 35,000V electric
field, they are accelerated and arrive at the anode with high velocity and considerable
kinetic energy. The intermediate electrodes (G1, G2, and G3) shape the electric field,
focusing the electrons properly onto the output layer, where the energetic electrons
strike the output phosphor and cause visible light to be emitted.
The Output Phosphor

The output phosphor (Fig. 9-5) is made of zinc cadmium sulfide doped with silver
(ZnCdS: Ag), which has a green (530 nm) emission spectrum. The ZnCdS phosphor particles are very small (1 to 2 mm), and the output phosphor is quite thin
(4to 8 mm), to preserve high spatial resolution. The anode is a very thin (0.2 mm)
coating of aluminum on the vacuum side of the output phosphor, which is electrically conductive to carry away the electrons once they deposit their kinetic energy
in the phosphor. Each electron causes the emission of approximately 1,000 light
photons from the output phosphor.
Light Absorber
Window
Output Phosphor
Anode
Electron
Trajectory
Figure 9-5 The output window of the II is shown. The electron strikes the very thin (0.2 mm or 200 nm)
anode and phosphor, and the impact of this high-velocity charged particle causes a burst of green light in the
phosphor, which is very thin (1 to 2 mm). The light exits the back side of the phosphor into the thick glass window. Light photons that hit the side of the window are scavenged by the black light absorber, but a high fraction of photons that reach the distal surface of the window will be focused using an optical lens and captured
by a digital camera. Light propagates by diffusion (multiple scatter interactions) in a turbid phosphor medium,
but in the clear glass window, light travels in straight lines until surfaces are reached, and then standard optics
principles (the physics of lenses) are used for focusing.
287
The image is much smaller at the output phosphor than it is at the input
phosphor, because the 15- to 35-cm-diameter input image is focused onto a 2.5-cm
(1 inch)-diameter circle. To preserve a resolution of 5 line pairs/mm at the input
plane, the output phosphor must deliver resolution in excess of 70 line pairs/mm.
This is why a very fine grain phosphor is required at the output phosphor. The reduction in image diameter also leads to light intensity amplification, since, for example,
the electrons produced on the 410 cm2 area of the 23-cm-diameter input phosphor
are concentrated onto the 5-cm2 output phosphor. By analogy, a magnifying glass
collects light over its lens surface area and can focus it onto a small spot, and in the
process, the light intensity is amplified enough to start a fire using only sunlight. The
so-called minification gain of an II is simply the ratio of the area of the input phosphor
to that of the output phosphor. The ratio of areas is equal to the square of the ratio of
diameters, so a 23-cm-diameter II (with a 2.5-cm-diameter output phosphor) has a
minification gain of (22.8/2.55)2 81, and in 17-cm mode it is 49.
The last stage in the II that the image signal passes through is the output window
(Fig.9-5). The output window is part of the vacuum enclosure and must be transparent to
the emission of light from the output phosphor. The output phosphor is coated directly onto
the output window. Some fraction of the light emitted by the output phosphor is reflected
inside the glass window. Such stray light reflecting inside the output window contributes to
veiling glare, which can reduce image contrast. This glare is reduced by using a thick (about
14mm) clear glass window, in which internally reflected light eventually strikes the side of
the window, which is coated with a black pigment to absorb the scattered light.
Optical Coupling to the Video Camera

A light-sensitive camera such as an analog vidicon or a solid-state CCD or CMOS
system is optically coupled to the output screen of the II and is used to relay the output image to a video monitor for viewing by the operator. The lenses used in fluoroscopy systems focus light emitted at the output phosphor onto the focal plane of the
video camera. The lens assembly includes an adjustable aperture, basically a small
hole positioned between individual lenses. Adjusting the size of the hole changes
how much light gets through the lens system. The f-number is inversely related to
the diameter of the hole (f 5 focal length/aperture diameter), but it is the area of the
hole that determines how much light gets through. Standard f-numbers increment by
multiples of 2: 1.0, 1.4, 2.0, 2.8, 4.0, 5.6, 8, 11, and 16. Changing the diameter of
the hole by a factor of 2 changes its area by a factor of 2, and thus, increasing the
f-number by one f-stop reduces the amount of light passing through the aperture by
a factor of 2. At f/16, the aperture is tiny, little light gets through, and the overall gain
(signal amplification of the light by minification gain and electron acceleration) of the
II and combined optics is reduced. At f/5.6, more light gets through, and the overall
gain of the II-optics subsystem is eight times higher than at f/16.
Adjustment of the lens aperture on II-based fluoroscopy systems markedly
changes the gain of the system, which has an important effect on its performance. By
lowering the gain of the II and associated optics by increasing the f-number, a higher
x-ray exposure rate will result, producing lower noise images and higher patient
dose. Increasing the gain reduces the x-ray exposure rate and lowers the dose, but
reduces image quality. This will be discussed more fully below (see Section 9.4).
IIs absorb x-ray photons and emit light photons, just as an intensifying screen
does, but the amplification of the II produces orders-of-magnitude more light. It is
this electronic gain that creates a very sensitive detector, which produces a viewable
output image using much less radiation compared to a radiographic detector.
288

B
picture data
mV
signal
time
electron beam
synch pulse
electron beam
TV target
signal plate
light
light
raster scan
TV Camera Target
monitor
phosphor
Video Monitor
Figure 9-6 A diagram illustrating the closed circuit analog TV system used in older fluoroscopy systems
is illustrated. At the TV camera (A), an electron beam is swept in raster fashion on the TV target (e.g. SbSO3).
The TV target is a photoconductor, whose electrical resistance is modulated by varying levels of light intensity.
In areas of more light, more of the electrons in the electron beam pass across the TV target and reach the
signal plate, producing a higher video signal in those lighter regions. The video signal (B) is a voltage versus
time waveform which is communicated electronically by the cable connecting the video camera to the video
monitor. Synchronization pulses are used to synchronize the raster scan pattern between the TV camera target
and the video monitor. Horizontal sync pulses (shown) cause the electron beam in the monitor to laterally
retrace and prepare for the next scan line. A vertical sync pulse (not shown) has different electrical characteristics and causes the electron beam in the video monitor to reset at the top of the screen. Inside the video
monitor (C),the electron beam is scanned in raster fashion, and the beam current is modulated by the video
signal. Higher beam current at a given location results in more light produced at that location by the monitor
phosphor (D). The raster scan on the monitor is done in synchrony with the scan of the TV target.
Video Cameras
The operation of a legacy fully-analog electron-beam (vidicon) video camera is shown
schematically in Figure 9.6. In the video camera, patterns of light (the image data) are
incident upon the TV target (Fig. 9.6A). The target is swept by a scanning electron
beam in a raster scan pattern. The TV target is made of a photoconductor, which has
high electrical resistance in the dark, but becomes less resistive as the light intensity
striking it increases. As the electron beam (electrical current) scans each location on
the target, the amount of current that crosses the TV target and reaches the signal
plate depends on the resistance of the target at each location, which in turn is related
to the local light intensity. Thus the electrical signal is modulated by the local variations in light intensity, and that is physically how the video system converts the optical image incident upon it into to an electronic signal. Note that this is a process of
sampling in time described as temporal sampling frequency with units of cycles/s,
with the unit hertz (Hz).
The video signal, which is represented as voltage versus time (Figure 9.6B), is
transferred by wire to the video monitor. Synchronization pulses are electronically
added to the signal to synchronize the raster scanning in the video camera and video
monitor. At the monitor, the raster scan pattern on the target is replicated by an electron beam scanning the monitor phosphor,. The video signal voltage modulates the
289
beam current in the monitor, which in turn modulates the luminance at that location
on the monitor (Figure 9.6C). The raster scan on the monitor is done in synchrony
with the scan of the TV target. (Figure 9.6D).
Vidicon video systems typically operate at 30 FPS in an interlaced scanning
mode to reduce flicker, the perception of the image flashing on and off. The human
eye-brain system can detect temporal fluctuations slower than about 47 images/s,
and therefore at 30 FPS, flicker would be perceptible. With interlaced systems, each
frame is composed of two fields (called odd and even fields, corresponding to every
other row in the raster, with the odd field starting at row 1, and the even field starting at row 2), and each field is refreshed at a rate of 60 FPS (although with only
half the information), which is fast enough to avoid the perception of flicker. One
drawback of interlaced scanning is the inability to uniformly read out a short pulsed
exposure (less than 33 ms), since the scanning electron beam occupies distributed
profile on the TV target, which causes some of the photoconductive charge patterns
to be removed from the alternate field when the other field is scanned. Because
light is not continuously incident on the target, the first field that is read will have
a much larger video signal (and therefore greater brightness on a monitor) than
the second field, and when interlaced, a bright/dark flicker pattern will be present.
(This was a common presentation problem with pulsed cardiac angiography studies
in legacysystems).
Charge coupled device (CCD) and complementary metal oxide semiconductor (CMOS) photosensitive cameras are solid-state electronic arrays that convert a
projection image into a digital image and then into a video signal. Such a device
uses photodiode semiconductor materials in an array to convert incident light intensity incident into corresponding electronic charge (electrons), and locally store the
charge for electronic readout. Both CCD and CMOS cameras have very small pixel
dimensions, on the order of 1020 micron (0.01 to 0.02 mm), incorporated into a
2.5 3 2.5 cm area that replaces an analog TV camera. The typical matrix comprises
1,000 3 1,000 or 2,000 3 2,000 detector elements, but there is a wide variety
and selection of sizes that a manufacturer might choose to include in a system.
Unlike analog photoconductive targets, which have a non-linear response to light,
CCD and CMOS cameras have a linear characteristic response over a very wide
brightness range. Both CCD and CMOS systems can integrate light produced from
intermittent pulsed exposures, or continuously read out fluoroscopic images in real
time (30 FPS). The CCD reads out the stored charge in a bucket brigade fashion,
as discussed in Chapter 7, causing locally stored electrons to be transferred from
one row to the next row, along defined columns, to eventually be deposited onto a
charge amplifier at the end of each column. The signals from the column amplifiers
are synchronized and convert each row of accumulated charge into a corresponding
digital value in the digital frame buffer. The CMOS device has independent detector
elements, each comprised of a storage capacitor and transistor switch. The signal
from a CMOS device is accessed, as is the signal from a TFT flat panel detector, as
discussed in chapter 7. In either output (CCD or CMOS) a digital projection image
is produced, and then converted to a video signal that is used to drive one or more
monitors in the fluoroscopy suite. Solid-state cameras are preferred because of their
stability, reproducibility, and linearity.
Characteristics Unique to Image Intensifier Systems

Brightness Gain
The brightness gain is the product of the electronic and minification gains of the
II. The electronic gain of an II is roughly about 50, and the minification gain is
290
v ariable depending on the electronic magnification (see magnification mode in

Section 9.5). For a 30-cm (12-inch) diameter II, the minification gain is 144, but
in 23-cm (9-inch) mode it is 81, and in 17-cm (7-inch) mode it is49. The overall
brightness gain, therefore, ranges from about 2,500 to 7,000. As the effective diameter (FOV) of the input phosphor decreases (increasing magnification), the brightness gain decreases.
Pincushion Distortion
Pincushion distortion was mentioned previously and is the result of projecting the
image with a curved input phosphor to the flat output phosphor (Fig. 9-4). Pincushion distortion worps the image by stretching the physical dimensions in the periphery of the image. Therefore, for improved accuracy with distance measurements, it is
best to position the desired anatomy in the central area of the FOV.
S Distortion
S distortion is a spatial warping of the image in an S shape through the image. This
type of distortion is usually subtle, if present, and is the result of stray magnetic fields
and the earths magnetic field affecting the electron trajectory from the cathode to the
anode inside the II. On fluoroscopic systems capable of rotation, the S distortion will
generally shift in the image during rotation due to the change in the IIs orientation
with respect to the earths magnetic field.
Flat Panel Detector Based Digital Fluoroscopy Systems

Flat panel detectors are comprised of thin film transistor (TFT) arrays of individual detector elements (dexels) that are packaged in a square or rectangular area. A
more complete discussion of TFT arrays can be found in Chapter 7. Both indirect
and direct x-ray conversion modes are used with TFT panels for fluoroscopy
applications. In both types of systems, each detector element has a capacitor, which
accumulates and stores the signal as an electrical charge, and a transistor which
serves as a switch. For indirect detection TFT systems, in which the absorbed x-rays
are initially converted to light in the adjacent layered phosphor, each dexel has a
transistor and a capacitor, in addition to a photodiode, which converts the x-ray
induced light from the phosphor into a corresponding charge. For direct detection fluoroscopy detectors, the semiconductor (selenium) produces x-ray induced
charge directly, which is collected under a voltage to ensure that the signal is captured within the same dexel as the x-ray absorption event. In either detector type,
while a frame is being acquired, the electrical charge proportional to the x-ray flux
incident on that dexel is accumulated and stored by the capacitor. Later, when the
image receptor is read, a signal is sent to all the transistors in a row of dexels and
the signals from the capacitors in the dexels of that row are simultaneously read as
the charge from the capacitor is sent down a corresponding drain line along each
column of the array to a charge amplifier. Reading the array discharges the capacitors and readies them for acquiring the next frame. This is performed row by row
until the full frame is read. Even in continuous fluoroscopic acquisitions, the charge
continues to be stored in each capacitor in a steady-state situation so that x-ray
information is always collected as the flat-panel image receptor provides real-time
fluoroscopic presentation of image data.
For fluoroscopic applications, the flat panel image receptor replaces the II,
lenses, and camera system, and directly records the real-time fluoroscopic image
sequence. The size of a detector element (dexel) in fluoroscopy is usually larger than
291
in radiography, and some flat panel systems have the ability to adjust the dexel size
by binning four dexels into one larger dexel. For example, four 100-mm detector
elements are grouped together (binned) electronically to form an effective 200 mm
3200 mm dexel. Such dual-use systems have dexels small enough for high-resolution
radiography (e.g., 100 to 150mm), but the dexels can be binned to provide a detector useful for fluoroscopy (e.g., 200 to 300 mm).
In some fluoroscopic flat-panel detectors, larger dexels (providing less spatial
resolution) are necessary for fluoroscopy because data transfer rate (bandwidth)
limitations restrict the amount of image data that can be displayed at 30 FPS. The
2 32 binning reduces the amount of data by a factor of four, making real-time
display feasible. In addition, the low doses per frame used in fluoroscopy results
in significant quantum noise, and ultimately the poor statistical integrity (low
SNR) limits the ability to see small objects. Therefore, larger dexel and poorer
spatial resolution are sometimes necessary for fluoroscopic operation. In other flat
panel detectors for fluoroscopy, the detector is read out always with full resolution, and to the extent that the number of pixels in the digital image excceds those
on the display monitor, real-time video processors interpolate and average the
information as displayed to the fluoroscopist. The effect of this averaging on the
video monitor results in an image presentation that appears similar to the binning
method, but in image playback mode the full image resolution can be viewed with
image zoom, unlike the systems that implement directly on the detector binning.
Flat panel receptors replace the II, optics, video camera, digital spot film device,
and cine camera in a much lighter and smaller package. A cover comprised of
carbon fiber (1 mm) instead of 1.5 mm Al in an II input screen reduces x-ray
attenuation and improves the quantum detection efficiency compared to the II
(Fig. 9-7).
Quantum Detection Efficiency (%)
100
20 cm patient
90
Flat panel
80
70
Image intensifier
60
50
40
50
60
70
80
kV
90
100
110
120
Figure 9-7 This figure compares the quantum detection performance as a function of x-ray beam energy
(kV of a polyenergetic spectrum hardened by a 20-cm-thick patient). In an II system, an x-ray has to penetrate
about 1.5 mm of aluminum before reaching the actual phosphor. A considerable fraction of incident photons
will be attenuated, reducing QDE, especially at lower kV values. The flat panel detector system also uses a CsI
phosphor, but only a carbon fiber cover is in the x-ray beam path, which acts as physical protection and as a
light shield to the detector.
292
9.4 Automatic Exposure Rate Control

The exposure rates in modern fluoroscopic systems are controlled automatically. The purpose of the automatic exposure rate control (AERC) circuit (formerly
referred to as Automatic Brightness Control ABC) is to keep the signal-to-noise
ratio (SNR) of the image constant when possible. It does this by regulating the
x-ray exposure rate incident on the input phosphor of the II or flat panel detector. When the system is panned from a region of low attenuation to one of greater
attenuation (e.g., thin to a thick region of soft tissue) of the patient, fewer x-rays
strike the detector. An AERC sensor (Fig.9-1) measures the x-ray intensity and
the AERC sends a signal to the x-ray generator to increase the x-ray exposure rate.
The AERC circuitry strives to keep the detected photon fluence used for each fluoroscopic frame at a constant level, keeping the SNR of the image approximately
constant regardless of the thickness of the patient. It also increases the exposure
rate when magnification modes are used, so that the detected photon fluence per
pixel is approximately constant as the effective pixel area on the displayed image
becomes larger. The AERC circuitry in the x-ray generator is capable of changing the mA and the kV in continuous fluoroscopy mode. For pulsed fluoroscopy
systems, the AERC circuitry may regulate the pulse width (the time duration) or
pulse height (the mA) of the x-ray pulses as well as kV (to increase or decrease
penetrability of the beam).
For large patients, the x-ray exposure rate may reach a regulatory limit, discussed later in this chapter, yet the image will still not be appropriately bright.
Some II-based fluoroscopic systems employ AERC circuitry, which can modulate
the lens aperture to increase the conversion efficiency (i.e., gain) of the system
when the exposure rate reaches regulatory limits. With current technology flatpanel detectors for fluoroscopy, the charge amplifier gain is fixed (although variable gain, avalanche detector technologies are in the development stage). When
x-ray exposure rates reach a maximum (e.g., 87 mGy/min (10 R/min) maximum
tabletop exposure rate) the visible noise in the image will be very apparent.
Special activation controls (175 mGy/min [20 R/min] tabletop exposure) can
be utilized in such circumstances to reduce excessive noise when absolutely
necessary.
For AERC, the x-ray generator changes the mA and the kV in a predetermined manner; however, this varies with different models of generators and is
often selectable on a given fluoroscopic system. How the mA and kV change
as a function of patient thickness has an important influence on the compromise between patient dose and image quality. When the fluoroscopist pans to
a thicker region of the patient, the AERC circuit requests more x-rays from the
generator. When the generator responds by increasing the kV, the subject contrast decreases, but the dose to the patient is kept low because more x-rays
penetrate the patient at higher kV. In situations where contrast is crucial (e.g.,
angiography), the generator can increase the mA instead of the kV; this preserves
subject contrast at the expense of higher patient dose. In practice, the mA and
kV are increased together, but the curve that describes this can be designed to
aggressively preserve subject contrast, or alternately, to favor a lower dose examination (Fig. 9-8). Many fluoroscopy systems have low dose, conventional,
or high contrast selections on the console, which select the different mA-kV
curves shown in (Fig. 9-8).
293
Figure 9-8 AERC circuitry adjusts the x-ray output in response to the thickness of the patient. X-ray output
can be increased for thick regions by increasing the mA, kV, or both. Increasing kV increases the dose rate but
contrast is reduced, whereas increasing mA raises the dose rate but preserves contrast. Different AERC curves
can be used on some systems (five are shown, typically fewer are available), which allow the user to select the
low-dose or high-contrast AERC logic, depending on the clinical application.
9.5 Fluoroscopy Modes of Operation

Modern x-ray generators and imaging systems are controlled by computers, resulting
in a great deal of flexibility regarding operational modes compared to the availability
in the 1990s and earlier. A brief description of various fluoroscopy modes, analog
and digital, is given below.
Continuous Fluoroscopy
Continuous fluoroscopy produces a continuous x-ray beam typically using 0.5 to
6 mA (depending on patient thickness and system gain). A camera displays the
image at 30 FPS, so that each fluoroscopic frame is displayed for 33 ms (1/30 s).
Any motion that occurs within the 33-ms acquisition blurs the fluoroscopic image.
This mode is the most basic approach to fluoroscopy acquisition, is typically used
on all analog systems, and was the standard operating mode for image-intensified
fluoroscopy up until the 1980s. The TV camera-based systems typically operated
with either 525 line or 1,000 (1023, 1024) line video output in an interlaced
scanning format. Modern fluoro systems may permit operation in this mode and
some very simple mobile mini-C-arm systems for extremity use only work in
this mode.
Variable Frame Rate Pulsed Fluoroscopy

In pulsed fluoroscopy, the x-ray generator produces a series of short x-ray pulses
(Fig.9-9). The pulsed fluoroscopy system can generate 30 pulses/s, but each pulse
can be very short in time. With pulsed fluoroscopy, the exposure time ranges from
about 3 to 10 ms instead of 33 ms, which reduces blurring from patient motion (e.g.,
pulsatile vessels and cardiac motion) in the image. Therefore, in fluoroscopic procedures where object motion is high (e.g., positioning catheters in highly pulsatile vessels), pulsed fluoroscopy offers better image quality at the same average dose rates.

Same area
mA
8 ms
30 FPS
mA
33 ms
30 FPS
mA
Pulsed
Fluoroscopy
Continuous
Fluoroscopy
294
15 FPS
Time
Figure 9-9 This figure illustrates three different pulse sequences for fluoroscopy. The top (a) is continuous
fluoroscopy, which was the standard operating mode for fluoroscopy for decades. The 33-ms frame integration time led to motion blurring in highly pulsatile regions of the patient. The middle sequence (b) shows a
pulsed fluoroscopy sequence at 30 FPS. Here, the pulse width is 8 ms instead of 33, while the mA is increased
by the factor 33/8, to deliver images of similar image quality (same dose) but with less motion blurring. The
bottom sequence (c) shows a 15-FPS sequence, which reduces not only the temporal sampling but also the
radiation dose by a factor of 2. To reduce dose, the fluoroscopist should use the lowest frame rate that is
practical for the clinical task at hand. The hatched areas shown in sequences a and b illustrate that the area of
each pulse is the product of the time (s) and mA and hence is the mAs per fluoroscopic frame.
During much of a fluoroscopic procedure, temporal resolution with high frame

rate is not required. For example, in a carotid angiography procedure, initially guiding
the catheter from the femoral artery access to the aortic arch might allow the use of
7.5 FPSreducing the radiation dose for that portion of the study to 25% (7.5/30).
Pulsed fluoroscopy at selectable frame rates (typically 30, 15, 7.5 and 3.75 FPS)
allows the fluoroscopist to reduce temporal resolution when it is not needed, sparing dose. At low frame rates, the fluoroscopic image would demonstrate intolerable
flicker if not displayed with continuous refresh during the time intervals between the
pulses. However, all modern fluoroscopy systems use a digital refresh memory, where
each image is captured digitally and is displayed continuously at the refresh rate of
the fluoroscopy display monitor. (e.g., 60 to 72 Hz or faster) until it is refreshed with
the next x-ray pulse and newer digital image. In this way, a slightly jerky temporal
progression of images may be seen by the human observer at low frame rates, but the
image does not suffer from overt on-and-off flicker that would occur if the displayed
image flickered off and on synchronously with the x-ray pulse.
Field of View and Magnification Modes

IIs are available in different sizes, commonly 23-, 30-, 35-, and 40-cm diameter FOV.
Large IIs are appropriate for GI/GU work, where it is useful to image the entire abdomen. For cardiac imaging, the 23-cm (9-inch) II is adequate for imaging the heart, and
its smaller size allows closer positioning in a typically biplane configuration and craniocaudad tilt without moving the center of the image receptor far from the patient.
In addition to the largest FOV, which is determined by the physical size of the II,
most IIs have several magnification modes, which yield higher spatial resolution for a
smaller field of view. Magnification is produced by changing the voltages applied to
the electrodes in the II, resulting in electron focusing from a smaller central area of the
295
Voltages applied
to electrodes
adjust focusing
ABC increases
X-ray fluence
Collimators
Adjust
No Magnification
With Magnification
Figure 9-10 No-magnification (left) and magnification (right) modes are illustrated. When the magnification mode is selected, the electrode voltages in the II are changed such that the electrons are focused to a smaller
FOV. The motorized collimator blades also constrict the x-ray field to the smaller region. The AERC senses a loss
of light immediately after switching to the magnification mode, and increases the exposurerate.
IIs input screen to the entire output phosphor. As the magnification factor increases,
a smaller area on the input of the II is seen (Fig. 9-10). When the magnification mode
is engaged, the x-ray beam collimator automatically adjusts to match the x-ray beam
dimensions to the smaller FOV.
As discussed previously, the brightness gain (and hence the overall gain) of the
II decreases as the magnification increases. The AERC circuitry compensates for the
lower signal by boosting the x-ray exposure rate. The increase in the exposure rate is
commonly equal to the ratio of FOV areas, although the user can deviate from this
algorithm to lower the radiation dose to the patient at the expense of greater quantum noise in the image. Consider a 30-cm II, which may have 23- and 18-cm magnification modes. Switching from the 30-cm mode to the 23-cm mode will increase the
x-ray exposure rate by a factor of (30/23)2 5 1.7, and going from the 30-cm FOV to
the 17-cm FOV will increase the exposure rate by a factor of (30/17)2 5 3.1. Imageintensified fluoroscopy systems use a sensor at the output of the II to drive the AERC,
so the overall gain of the II and optics systems drives the AERC feedback operation.
Flat panel fluoroscopy systems can provide similar exposure rate adjustment as a
function of field size as IIs; however, the exposure rate is controlled by system logic
and is explicitly calibrated into the fluoroscopy system. In designing fluoroscopy
systems using flat panel detector systems, it is logical that the same approximate dose
rate-versus-FOV performance as with II systems should be used, since that is what
users are familiar with.
In comparing image quality between imaging intensifiers and flat panel detector
systems, although the rectangular format, compact design, and intrinsic digital attributes of flat panel systems enhance functionality, the electronic gain circuitry in the II
produces images with very little added electronic noise. Flat panel systems, however,
have amplifier systems that work at the very limit of what compact digital electronics
can deliver, and consequently, there is appreciably more electronic noise in most flat
296
panel fluoroscopy systems. Thus, flat panel fluoroscopy detectors are calibrated to
use slightly higher exposures per frame compared to IIs of similar area. Electronic
noise is far less of an issue in flat panel detectors when they are used in higher dose
fluorographic and radiographic acquisition modes.
To minimise patient dose, the fluoroscopist should use the least magnification
and smallest collimation area that facilitates the diagnostic task at hand.
Frame Averaging
Fluoroscopy systems provide excellent temporal resolution, a feature that is the basis
of their clinical utility. However, fluoroscopy images are also relatively noisy, and
under certain circumstances it is appropriate and beneficial to (reduce) temporal
resolution for lower quantum noise. This can be accomplished by averaging a series
of images, as shown in Figure 9-11. Appreciable frame averaging can cause noticeable image lag with reduced temporal resolution. The compromise depends on the
specific fluoroscopic application and the preferences of the user. Aggressive use of
frame averaging can provide lower dose imaging in many circumstances. Mobile
fluoroscopy systems use x-ray tubes with limited output, and consequently temporal
frame averaging is a common feature.
C
Perturbation
Video Signal at one pixel
1.1
1.0
= 1.0
0.9
= 0.5
0.8
= 0.3
0.7
= 0.1
0.6
0.5
0.4
0
10
20
30
40
50
60
70
80
90
100
Fluoroscopic Frame Number
Figure 9-11 A. This image shows the trade-off between temporal blurring and noise. Frame A integrates
five frames, and there is minimal blurring of the moving central object, but the image is overall noisier. Frame
B shows 10-frame blurring, with a considerably broader central object but lower noise in the static areas of
the image. Frame C (20-frame averaging) shows dramatic blurring of the central, moving object but enjoys
a very high SNR in the static regions of the image. D. This figure shows the influence of different recursive
filters for temporal averaging. A parameter (a) setting of 1.0 results in no temporal filtering, and a noisy trace
at this one (arbitrary) pixel is seen. As the value of a is lowered, the amount of temporal lag is increased and
the trace becomes smoother. However, the rounded leading edges (at t 5 1 and t 5 40) show the increased
lag that occurs in the signal.
297
Different temporal averaging algorithms can be used, but a common approach to

frame averaging is called recursive filtering. In this approach, the image just acquired,
In, is added together with the last displayed image (In1) using
I displayed = I n + (1 )I (n1)
[9-1]
The parameter a ranges from 0 to 1; as a is reduced, the contribution of the

current image (In) is reduced, the contribution of the previous displayed image is
enhanced, and the amount of lag is increased. As a is increased, the amount of lag is
reduced, and at a 5 1, no lag occurs. The current displayed image Idisplayed becomes the
In1 image for the next frame, and the weighted image data from an entire sequence of
images (not just two images) is included (Fig. 9-11). For example, for a 5 0.3, if the
x-rays are abruptly turned off, the signal in the displayed image would be less than
1% after 17 blank (no input) frames at 30 FPS, or about 0.56 s.
Last-Frame-Hold, Last-Sequence-Display
When the fluoroscopist steps off the pedal and deactivates fluoroscopy, rather than
seeing a blank monitor, last-frame-hold continuously displays the last acquired image
or set of images on the fluoroscopy monitor. Last-frame-hold is a standard feature on
modern fluoroscopy systems, and is achieved by continuously digitizing images in
real time and temporarily storing them in a digital video frame memory. when the
fluoro pedal is released, are displayed the last bright image or images on the monitor. Rather than orienting oneself with patient anatomy while the fluoroscopy pedal
is depressed and x-rays are on, last-frame-hold allows the fluoroscopist to examine
the image on the monitor for as long as necessary, using no additional radiation. A
playback of the last sequence of images, programmable over several seconds, is available on many advanced fluoro systems with larger video buffer memories, and can
be useful for complicated procedures. Last-frame-hold is a necessary feature for dose
reduction, especially at training institutions.
Road Mapping
Road mapping is a software- and video-enhanced variant of the last-frame-hold
feature and is useful for angiography procedures. Two different approaches to road
mapping can be found on fluoroscopic systems. Some systems employ side-by-side
video monitors, and allow the fluoroscopist to capture an image (usually with a
small volume of injected contrast media), which is then shown on the monitor next
to the live fluoroscopy monitor. In this way, the path of the vessel can be seen on
one monitor, while the angiographer advances the catheter viewing in real-time on
the other monitor. Another approach to road mapping is to capture a contrast injection image or subtracted image, and use this as an overlay onto the live fluoroscopy
monitor. In this way, the angiographer has the vascular road map superimposed on
the fluoroscopy image and can orient the guidewire or catheter tip to negotiate the
patients vascular anatomy. Road mapping is useful for advancing catheters through
tortuous vessels.
C-arm CT
Some modern C-arm fluoroscopic systems for angiography have computed tomography (CT) acquisition modes (Fig. 9-12). The C-arm is capable of motorized rotation
of about 220 degrees around the patient, which allows two-dimensional projection
298
Figure 9-12 C-arm CT on an angiographic system makes use of a motorized rotating C-arm, and acquires
image data over about a 220-degree arc around the patient.
images to be acquired at many different angles around the patient. This data is then
reconstructed using cone beam reconstruction algorithm. Images from such a conebeam CT acquisition have poorer image quality compared to a conventional CT scanner, but can be of value in applications such as angiography. Details on CT and cone
beam reconstruction are discussed in Chapter 10. The fidelity of the images is not
as good as with whole body CT scanners, because the rotational axis on the C-arm
system is not as rigid, and the cone beam geometry results in the detection of much
higher scatter levels compared to whole body CT systems. Nevertheless, C-arm CT
provides tomographic capability during interventional procedures, and this provides
interventionalists an important additional tool in validating the success of the procedure. Dose levels during these procedures are high, and so physicians should use
this acquisition mode sparingly.
9.6 Image Quality in Fluoroscopy
Spatial Resolution
For routine quality control testing in fluoroscopy, a visual assessment of a resolution test object imaged with minimal geometric magnification is commonly used
to assess spatial resolution (Fig. 9-13). The intrinsic limiting resolution of modern fluoroscopy detectors is quite high, ranging from 3 to 5 cycles/mm; however,
the limiting resolution of the entire system is most often governed by the video
device and the effective sampling matrix size across the FOV, the latter determining the pixel dimension in the displayed image. A 23 cm FOV II/TV system with a
1024 3 1024 matrix will have a pixel dimension of 230 mm / 1024 5 0.22 mm,
which translates to about 2.2 cycles/mm (Table 9-1). Higher magnification modes
provide better resolution; for example, an 11-cm mode may have a limiting resolution up to 3.5 cycles/mm.
Flat panel detectors for fluoroscopy are designed to deliver similar or better
spatial resolution as image-intensifier systems. Arrays with detector elements having similar dimensions as II/TV systems with or without pixel binning are being
used. In many systems, the full sampling resolution of the flat panel is maintained,
299
5
1.5
Resolution (lp / mm)
1.5
0.5
2.0
0.7
0.7
0.6
2.5
3.0
3.0
Flat panel system
0.5
2.0
0.6
2.5
0.8
3.5
0.8
3.5
4.0
0.9
4.0
0.9
5.0
1.0
5.0
1.0
lmage intensifier
1
0
10
15
20
25
30
35
40
45
50
Flat Panel Field Size (cm)

Figure 9-13 The spatial resolution in fluoroscopy for an II based fluoro system and a flat panel system is
shown. As the field size decreases, magnification increases, and the spatial resolution increases commensurately. The resolution template is shown before (left) and after (right) image blurring.
independent of the displayed FOV. Spatial resolution in this instance is limited by

the monitors display matrix size and the software which interpretates and scales the
full FOV image onto the monitor. With image zoom and panning, however, the full
inherent spatial resolution of the acquired image can be obtained, regardless of the
mode of operation. For systems with detector element binning, the intrinsic resolution is reduced by the binning factor (e.g., 2 3 2 binning reduces resolution by a
factor of 2.
The theoretical intrinsic resolution of fluoroscopy cannot exceed the Nyquist
frequency, FN 5 1/2 , where is the size of a detector element. Thus, for a system with a 30-cm FOV and 1,024 3 1,024 imaging matrix, 5 300mm/ 1,024
50.29mm; FN 5 1/(0.2932) 5 1.7 line pairs/mm.
The Nyquist frequency describes the maximum possible resolution in a digitally
sampled imaging system; Overall spatial resolution is also determined by the calibration of electronic and optical focusing of the II/TV system, pincushion distortion of
the II at the periphery of the image, and the size of the focal spot.
Geometric magnification, when used in conjunction with a small focal spot, is
sometimes useful in overcoming the spatial resolution limitations of a fluoroscopy
system, and is achieved by moving the detector away from the patient. Downsides
to geometric magnification are the large increase in dose nesessary to address the
increase in source to detector distance and the reduced anatomical coverage for a
given detector field at view.
TABLE 9-1TYPICAL LIMITING SPATIAL
RESOLUTION
IN FLUOROSCOPY
FOV: cm (inches)
525 LINE VIDEO

LINE PAIRS/mm
1023 LINE VIDEO

LINE PAIRS/mm
flat panel 0.157

mm LINE PAIRS/mm
14 (5.5)
2.7
1.4
3.2
20 (7.9)
2.0
1.0
2.8
27 (10.6)
1.6
0.7
2.5
40 (15.7)
1.3
0.5
1.8
300
Figure 9-14 Contrast resolution in fluoroscopy is essentially a measure of image noise. Many different types of phantoms (and testing procedures) are used, but shown here is a step wedge with holes
of different depths drilled into each of the steps. The thicker step attenuates more x-rays, resulting in a
noisier image. The noise differences between steps in this example were amplified for the purposes of this
demonstration.
Contrast Resolution
The contrast resolution of fluoroscopy compared to radiography is low, chiefly due
to the low SNR. Contrast resolution is usually measured subjectively by viewing
contrast-detail phantoms under fluoroscopic imaging conditions (Fig. 9-14). When
higher exposure rates are used contrast resolution increases, but dose to the patient
also increases. The use of exposure rates consistent with the image quality needs
of the fluoroscopic examination is an appropriate guiding principle. Fluoroscopic
systems with different dose settings (selectable at the console) allow the user flexibility
from patient to patient to adjust the compromise between contrast resolution and
patientdose.
Scattered radiation from the patient also has a significant impact on contrast
resolution, and therefore the use of an antiscatter grid in front of the fluoroscopy
detector is employed in most situations other than pediatric and extremity imaging. The antiscatter grid selectively removes contrast-degrading scatter and preferentially allows primary radiation to pass, improving contrast sensitivity. However,
the use of a grid causes a 2 to 4 times increase in radiation dose to the patient
because of compensation by AERC due to the attenuation of x-rays by the grid
(both scatter and primary). Grids with grid ratios of 6:1 to 12:1 and short (100 cm
or less) focal distances are used in fluoroscopic examinations. In some systems,
the grid can be moved in and out of the beam by the fluoroscopist. It is important
that this feature be employed, particularly in the case of smaller-size pediatric
patients.
Temporal Resolution
The excellent temporal resolution and its ability to provide real-time images
are the advantages of fluoroscopy in comparison to radiography. As mentioned
previously, temporal blurring is typically called image lag. Lag implies that a
fraction of the image data from one frame carries over into the next frame, and
this can be intentionally affected using algorithms such as recursive filtering
(Equation 9-1). By blurring together several temporal frames of image data, a
higher SNR is realized, because the x-ray photons from the several images are
301
combined into one image. In addition to the lag of the fluoroscopic system, the
photoreceptors in the human eye produce a lag of about 0.2 s. At 30 FPS, this
means that approximately six frames are averaged together by the lag of the
human eye-brain system. An image with six times the x-ray photons increases
the SNR by 6 2.5 times, and thus the contrast resolution improves at the
expense of temporal resolution. Whereas some lag is usually beneficial for unrecorded fluoroscopic viewing, for recording dynamic events such as in digital
subtraction angiography, lag is undesirable and recursive filtering modes are
disabled. The significantly higher exposure per frame (and higher SNR) in DSA
negate the need for intentional lag.
9.7 Fluoroscopy Suites

The clinical application strongly influences the type of fluoroscopic equipment
that is needed. Although smaller facilities may use one fluoroscopic system for a
wide variety of procedures, larger hospitals have several fluoroscopic suites dedicated to specific applications, such as GI/GU, cystography, peripheral vascular
and cardiac angiography, cardiac electrophysiology, and neurovascular imaging
procedures.
Genitourinary
In GI fluoroscopic applications, the radiographic/fluoroscopic room (commonly
referred to as an R and F room) has a large table that can be rotated from horizontal
to vertical to put the patient in a head-down (Trendelenburg) or head-up position
(Fig.9-15A ). The detector is typically mounted over the table, with the x-ray tube
under the table. The table has a large metal base that hides the fluoroscopic x-ray
tube and provides radiation shielding. For radiography, an overhead x-ray tube is
mounted on a ceiling crane in the room. The table may have a Bucky tray for use with
screen-film, CR and portable DR cassette detectors. Typically the cassette holder has
a moving antiscatter grid and has the capability of being aligned with the overhead
x-ray tube. Alternately, a DR detector system can be built directly into the table.
Remote Fluoroscopy Rooms

A remote fluoroscopy room is designed for remote operation by the physician in
charge of the examination (e.g., a radiologist or urologist). The fluoroscopic system is motorized and is controlled by the operator sitting in a shielded control
booth. In addition to the normal operation of the detector (pan, zoom, up, down,
magnification), a remote room typically has a motorized compression paddle that is
used to remotely palpate the patient, to manipulate contrast agent in the upper or
lower GI tract. Remote rooms use a reverse geometry, with the x-ray tube above the
table and the detector system below the table. These systems reduce the radiation
exposure to the physician operating it and reduce fatigue by allowing the procedure
to be performed sitting down and without a lead apron on. However, because of
the remote operation, these rooms require more operator experience. Remote rooms
with over-table x-ray tubes should not be used with the operator in the room, as a
GI/GU would be used. The x-ray scatter from the patient is most intense on the
entrant side of the patient and, with the over-table x-ray tube geometry, that means
302
Figure 9-15 A. An II-based fluoroscopy system for GI/GU applications is shown. This system has a foot
board to allow the patient to be tilted upright in a standing position, in order to move barium contrast
agent in the GI track. B. An II-based angiography system with a floating table is shown. This is a body
angiography system with a relatively large (40 cm) II. C. A flat panelbased, biplane, cardiac catheterization laboratory is shown. The smaller format of the flat panel detectors permits tighter positioning.
Notice the lead drape system mounted on the table for radiation protection. D. Electrophysiology lab
with permanent magnet and steerable catheter system that allows the operator to map cardiac functions
remotely, outside theroom.
that the highest scattered radiation levels will strike a person standing next to the
table in his or her upper body and face.
Peripheral Angiography Suites

In the angiographic setting, the table does not rotate but rather floatsit allows
the patient to be moved from side to side and from head to toe (Fig. 9-15B). The
fluoroscopy imaging chain (the x-ray tube and detector) can also be panned around
the stationary patient. The fluoroscopic system is typically mounted on a C-arm
apparatus, which can be rotated and skewed, and these motions provide considerable flexibility in providing standard posteroanterior, lateral, and oblique projections. Because of the routine use of iodinated contrast media, power injectors (for
injecting intravenous or intra-arterial iodinated contrast material) are often table- or
303
ceiling-mounted in angiographic suites. For peripheral angiography and body work,

the detector dimensions run from 30 to 44 cm. For neuroangiography rooms, 23- to
30-cm detector dimensions are commonly used.
Cardiology Catheterization Suite

The cardiac catheterization suite is very similar to an angiography suite, but typically
smaller (23 cm) flat panel detectors are used (Fig. 9-15C). The smaller detectors
permit more tilt in the cranial caudal direction, as is typical in cardiac imaging. High
frame rate (15 to 60 FPS) pulsed x-ray cineradiography operation is mandatory to
capture injected contrast medium as it moves through the vessels in the rapidly beating heart. Because multiple cine acquisitions are acquired, many cardiac catheterization systems are biplane (see below) to reduce the amount of contrast material that
must be administered.
Cardiology Electrophysiology Laboratory

Interventional cardiac electrophysiology (EP) studies chiefly use fluoroscopy to guide
pacing and recording electrodes that are introduced into the heart to stimulate and
measure the electrical properties of various cardiac control functions. These studies can sometimes take hours, and therefore it is important that the interventional
system be tuned for very low fluoroscopy dose rates by using heavy beam filtration,
very low pulsed fluoroscopy frame rates (3.75 or 7.5 FPS), and high sensitivity
fluoroscopy detectors. Advanced EP laboratories use remote fluoroscopy capabilities coupled with high-field strength magnets that can steer intracardiac catheters to
target locations in the heart with excellent accuracy. (Fig 9.15D)
Biplane Angiographic Systems

Biplane angiographic systems are systems with one patient table but with two complete imaging chains; there are two generators, two fluoroscopic detectors, and
two x-ray tubes (Fig. 9-15C). Two complete x-ray tube-detector systems are used
to simultaneously record two approximately orthogonal views following a single
injection of contrast material. The additional complexity and cost of a biplane system is to reduce the amount of iodinated contrast agent that is injected, to avoid or
reduce contrast reactions, and to allow more vessels to be evaluated. Two imaging
chains allow two orthogonal (or oblique) views of the area of interest to be acquired
simultaneously with one contrast injection. During biplane angiography, the x-ray
pulse sequence of each plane is staggered so that scattered radiation from one imaging plane is not imaged by the other plane. Biplane suites usually allow one of the
imaging chains to be swung out of the way when single plane operation is desired.
Mobile FluoroscopyC Arms

Mobile fluoroscopy systems are C-arm devices, and are used frequently in operating
rooms and intensive care units. These systems plug into a standard wall plug for
power, and therefore are relatively low power. Most C-arm systems use 15- to 23-cm
detector systems. The typical system has two movable units, one is the C-arm and
base itself, and the other unit (which can be connected by cables to the C-arm unit)
contains the fluoroscopic controls and display monitors.
304
9.8 Radiation Dose
Patient Dose Rate

The US Food and Drug Administration (FDA) regulates the manufacture of medical equipment and fluoroscopic equipment in the u.s. The maximum permissible
entrance exposure rate to the patient for normal fluoroscopy is 87.3 mGy/m or (10
R/m). For specially activated fluoroscopy, the maximum exposure rate allowable is
175 mGy/m (20 R/min). These dose rates limits are assessed at specified positions
for systems in which the x-ray tube is below the patient table, the measurement
position is one cm above the table; for C-arm fluoroscopes, this position is 30 cm
from the image receptor toward the x-ray source along the central axis of the beam.
Most states impose similar regulations on the users of fluoroscopes in their states.
However, neither the FDA nor the states provide dose limits for image recording.
Typical entrance exposure rates for fluoroscopic imaging are about 8.7 to 17 mGy/m
(1 to 2 R/min) for thin body parts and 26 to 44mGy/m (3 to 5 R/min) for the average
patient. The dose rate may be much larger for oblique and lateral projections and for
obese patients. The entrance exposure rate to the skin of a patient can also substantially exceed these values if the skin of the patient is much closer to the x-ray source
than the dose reference point.
Dose rates are evaluated in fluoroscopy by using a tissue-equivalent phantom
made of a material such as polymethyl methacrylate (PMMA, also known as Lucite
or Perspex) placed in the field, with an ionization chamber positioned in front of
it to measure the entrance skin dose (Fig. 9-16). Dose rates are measured with the
Figure 9-16 This figure shows the procedure for measuring the entrance dose rates on a fluoroscopy
system. The exposure meter is positioned under the PMMA using blocks, and different thicknesses of PMMA
are used to assess the dose rate over a range of magnification factors and operating modes (different fluoro
schemes, cine, DSA, etc.) of the system. To determine the maximum tabletop exposure rate, a sheet of lead is
positioned in the back of the PMMAthis drives the system to maximum output and allows the measurement
of maximum exposure rates.
305
fluoroscopic system operating in all typical modes (e.g., magnification modes, low
dose/high contrast, cine, DSA, etc), with the antiscatter grid in. Dose rates also are
measured using different thicknesses of PMMA to determine the entrance dose as a
function of patient thickness, for example, using 10, 20, and 30 cm of PMMA. To
determine the maximum exposure rate, a sheet of lead is placed between the PMMA
and the detector system to attenuate the x-ray beam and drive the AERC system to its
maximum output. The geometry shown in Figure 9-16 includes the measurement of
backscatter, because the ionization chamber is very close to the PMMA attenuator.
The overall sensitivity of the fluoroscopic system in a particular operating mode
can be characterized directly, and this approach is used when service personnel calibrate the fluoroscopic systems during installation and service. The grid is removed,
and all objects are removed from the FOV, including the table. A sheet of approximately 0.5 - 1.2 mm of copper is taped to the x-ray tube port to provide some beam
hardening so that the x-ray beam spectrum striking the detector is roughly similar
that used during patient imaging. An ionization chamber is placed over the detector, and the exposure rate is measured at the surface of the detector. Typicalentrance
kerma rates for fluoroscopic detectors range from 8.7 nGy (1 mR) to 44 nGy
(5 mR) per frame. Figure 9-17 shows the dose rate at the detector, as a function of
field size, for several fluoroscopic systems. For the II, these data are simply a computed curve with 1/x2 performance normalized to 17 nGy/frame for a 23-cm-diameter II. The other curves shown correspond to exposure rates measured on flat panel
fluoroscopic systems. These data fit curves with exposure rate dependencies of 1/x1.9,
1/x1.7, and 1/x1.5, respectively, where x is the side dimension of the rectangular FOV.
Dose to Personnel
Occupational exposures of physicians, nurses, technologists, and other personnel
who routinely work in fluoroscopic suites can be high. As a rule of thumb, standing
1 m from the patient, the fluoroscopist receives from scattered radiation (on the
140
120
nGy / frame
100
80
60
40
20
0
10
15
20
25
30
35
40
Field Size (cm)

Figure 9-17 The dose to the detector per frame is shown for several different fluoroscopic systems. The
curve with diamond symbols shows an II-based fluoroscopy system, while the other curves correspond to different flat panel imaging systems. It is seen that the flat panel system response, which is calibrated into the
system, closely matches that of the II.
306
outside of his or her apron) approximately 1/1,000 of the exposure incident upon
the patient. To reduce radiation exposure in fluoroscopy, everyone in the fluoroscopy
suite should wear radiation attenuating aprons, always. Movable lead shields should
be available for additional protection to staff members observing or participating
in the procedure. The dose to personnel is reduced when the dose to the patient is
reduced, so measures such as reducing the total fluoroscopy time, using variable rate
pulsed fluoroscopy at low pulse rates, and aggressive beam collimation are beneficial
to everyone. Stepping back from the radiation field helps reduce dose to the fluoroscopist and other staff, particularly during image recording when dose rates are typically higher. In addition to the protective apron, lead eyeglasses, a thyroid shield, and
the use of ceiling-mounted and table-mounted radiation shields can substantially
reduce the radiation doses to personnel in the fluoroscopy suite.
Stray radiation, comprised of both leakage and scatter radiation, has a distinctly
non-uniform distribution, chiefly due to backscattered radiation from the entrant
beam surface. A factor of 10 or more scatter difference exists between the entry and
exit locations on the patient, as shown in Figure 9-18.
Dose Monitoring
Patient Dose Monitoring
Active monitoring of doses to patients by fluoroscopy systems is routinely performed during exams that can lead to large skin dose levels, such as interventional neuroangiography and cardiology procedures. The purposes of patient
dose monitoring are to assess risks to individual patients and for quality assurance. With large radiation doses delivered in a short time for these types of
procedures, deterministic effects, including skin injury, hair loss, and cataract
formation, can occur (See Chapter 20 on Radiation Biology). One would like
to know the peak skin dose or have a map of skin doses, as in some cases the
x-ray tube and detector move over a considerable area of the body (e.g., cardiac
catheterization, while in other procedures, the x-ray tube movement is limited
(e.g., interventional n
euroangiography).
Dose monitoring may make use of ion chambers placed over the x-ray tube
and collimator assembly, as illustrated in Figure 9-1. These systems have various
acronyms such as kerma-area-product (KAP), roentgen-area-product, and dosearea-product meters. These systems are large ion chambers that are located after
the collimator, and so the meter does not indicate exposure or kerma rate, because
the entire chamber is not exposed as the collimator changes dimensions during
fluoroscopy. Consequently, KAP meters typically report output in the units of mGycm2. If field area in cm2 is known for a particular exposure (e.g., at the skin surface,
then the incident accumulated air kerma in mGy at that point can be calculated
by simply dividing the KAP measurement by the area. For the same exposure rate,
the KAP will give a larger value with a larger collimator setting. Since the collimated beam dimensions, patient position, and x-ray technique factors are constantly changing during most fluoroscopy and examinations, the KAP reading is
useful metric of accumulated dose, especially to the skin. This in turn can be useful
in managing the followup of patients who may reach dose levels that can cause
deterministic effects.
Active dose monitoring is important in interventional procedure rooms (typically
cardiac and neuroangiography suites) to alert physicians when skin dose levels are
getting so high as to potentially cause erythema. It is typically the responsibility of
the x-ray technologist or other personnel to monitor these dose levels during the
307
0.25
50 cm
8.0
4.0
2.0
1.0
0.5
0.25
mGy / h
90 degree LAO projection, 100 cm
0.25
50 cm
4.0
2.0
1.0
0.5
mGy/h
60 degree LAO projection, 100 cm
Figure 9-18 Stray (scatter plus leakage) radiation distributions surrounding the patient are not uniformly
distributed, as shown for three common projections used in cardiac catheterization procedures. A. 90 degree
LAO projection; B. 60 degree LAO projection.
308
1.0
0.25
0.5
50 cm
0.5
0.25
mGy / h
30 degree RAO projection, 100 cm
Figure 9-18 (Continued) C. 30 degree RAO projection. Isokerma maps at the plane of the tabletop are for
one manufacturers equipment and for a given patient size. These do not represent the stray radiation levels for
a general situation. The intent is to depict the factor of 10 and more difference between stray radiation between
the entry port and exit port of the beam on the patient for different projections, where backscatter predominates
the distribution. Figures are adapted from NCRP report 168, Radiation Dose Management for Fluoroscopically
Guided Interventional Procedures, The National Council on Radiation Protection and Measurements, Bethesda,
MD, 2011. LAO: Left Anterior Oblique; RAO: Right Anterior Oblique.
procedure. A procedural pause is required by some institutions when the integrated

skin dose reaches predefined levels, in order to evaluate the benefits and risk of continuing with the procedure.
Fluoroscopy systems manufactured on or after June 10, 2006 are required to
display the cumulative air kerma (commonly in the units of mGy) at a reference
point. For a C-arm fluoroscope, this reference point is on the beam axis, 15 cm from
the isocenter, toward the x-ray tube. The cumulative air kerma provides an actual
dose value that will have more meaning to most users. To avoid the additional cost
of active dose monitors, it is also possible (with proper calibration) to compute the
cumulative air kerma from the mA, time, kV, and SID used during the fluoroscopic
procedureno physical dose monitoring is necessary.
The cumulative air kerma may significantly over or underestimate the actual
peak skin dose for three reasons: (1) The cumulative air kerma does not account for
backscattered radiation. Due to back scatter, the actual skin dose will exceed the air
kerma at that location by a factor of about 1.25 to 1.5, depending upon the kV, beam
filtration, and x-ray field size. (2) The x-ray beam may be incident upon a single area
on the skin, or may be incident on multiple areas. If these areas do not overlap, the
dose is spread over the multiple areas. (3) The skin may closer or farther from the
x-ray tube than the reference point at which the cumulative air kerma is specified.
If the skin is much closer to the x-ray tube, the skin dose may greatly exceed the
cumulative air kerma.
309
Dosimeters are available that may be placed on a patients skin. Some are read after
the procedure and some give readings of cumulative dose during the procedure.
Monitoring of Doses to Personnel

Personnel who are routinely involved in interventional fluoroscopic procedures
usually wear personal dosimeters to measure their exposures. A number of technologies such as thermoluminescent dosimeters, film, and other dose integrating materials can be used; these are discussed in Chapter 21. Doses are reported
monthly or quarterly, to ensure that these individuals do not exceed dose limits
defined by the institution or state and to assess the effectiveness of dose-reduction
measures. It is customary to wear a single dosimeter at the collar level, in front
of the protective apron. Asecond dosimeter may be worn on the body under the
apron. Methods are available to estimate the effective doses from the readings
from the dosimeters (NCRP Report No. 122). A finger dosimeter may be worn to
monitor the dose to the hand likely to receive the largest exposure. Furthermore,
a self-reading dosimeter may be worn on the collar outside the apron to monitor
the dose from individual procedures.
Dose Reduction
Actively employing dose reduction techniques has immediate benefit to the patient
on the fluoroscopic table but also benefits the physicians and other personnel
who spend considerable time in the fluoroscopic suite. Techniques to minimize
the dose include heavy x-ray beam filtration (e.g., 0.2-mm copper), aggressive use
of low frame rate pulsed fluoroscopy, and use of low-dose (higher kV, lower mA)
AERCoptions.
The x-ray tube should be kept as far from the patients skin as possible. Aggressive beam collimation should be used. Although collimation does not reduce the
skin dose, it does limit the area irradiated, reduces the effective dose to the patient,
improves image quality by reducing the amount of scattered radiation, and reduces
dose to staff in the room. Removing the antiscatter grid is a selection on some fluoroscopy systems, and should be chosen when imaging patients of smaller size, particularly young pediatric patients and extremities. Using the lowest magnification mode
that is clinically acceptable reduces the higher dose that accompanies these modes.
Patient positioning is important: when possible, ensure that the patients arms are out
of the beam when using lateral projections; when the patient is supine, keep the x-ray
tube under the table to spare dose to the lenses of the eye and breast tissue in women;
and if possible, use dose spreading by varying the location of the x-ray entrance field
on the patient. The most effective way to reduce patient dose during fluoroscopy is
to use less fluoroscopy time. The fluoroscopist should learn to use fluoroscopic imaging sparingly, only to view motion in the image, and should release the fluoroscopy
pedal frequently. The last-frame-hold feature helps to reduce fluoroscopy time. Radiation dose during fluoroscopy procedures is not only from fluoroscopy per se, but
also from radiographic acquisitions: DSA sequences in the angiography suite, digital
spot images in the GI/GU fluoroscopy suite, and cineradiography runs in the cardiac
catheterization laboratory. Reducing the number of radiographic images, consistent
with the diagnostic requirement of the examination, is an excellent way of reducing patient dose. During cine and DSA radiographic acquisitions, when possible, all
personnel should stand behind portable lead shields or other radiation barriers in the
room, as these provide more protection than just a lead apron. A summary of how
changes in these operational factors affect image quality and radiation dose to the
patient and staff is given in Table 9-2.
310
TABLE 9-2Summary of operational factors that affect image
quality and radiation dose to the patient and staff

Effect on Image Quality & Radiation Dose
Radiation dose
to the patient
Radiation dose
to the staff
Worse (increased
scatter fraction)
Higher
Higher
Increase in tube current (mA)

with constant kV (i.e., AERC off)
Better (lower
image noise)
Higher
Higher
Increase in tube potential (kV)

with AERC active
Soft tissue: Better

(lower noise)
Bone & contrast
material: Worse:
(decreased subject
contrast)
Lower
Lower
Increase in tube filtration with

AERC active
Little change
Lower
Lower
Increase in source to skin

distance
Slightly better
Lower
Little change
Increase in Skin to image

receptor Distance
Slightly better
(less scatter)
Higher
Higher
Increase in Magnification
Factor
Better (improved
spatial resolution)
Higher
Higher
Increase in Collimator opening
Worse (increased
scatter fraction)
Little change
(however higher
integral and
effective dose)
Higher
Increase Beam on time
No effect
Higher
Higher
Increase in pulsed fluroscopy

frame rate
Better (improved
temporal resolution)
Higher
Higher
Grid is used
Better (decreased
scatter fraction)
Higher
Higher
Image recording modes (cine,

DSA, radiographic)
Better (lower noise,

higher resolution)
Higher
Higher
Operational Change
Image Quality
Increase in patient size
hen kV is increased with AERC activated, the system decreases the mA to maintain a constant signal level at
W
the image receptor.
2
Fixed source to skin distance.
1
Dose Recording and Patient Follow-up

Records of doses to the patient from each procedure should be kept for quality assurance purposes. Records of doses may also be entered into the patients medical record;
these doses should always be recorded when they are sufficiently large to potentially
cause tissue reaction. In high dose procedures, the stored information should include
the estimated skin doses and their locations on the patients skin.
If the estimated skin doses are large enough to cause skin injury, arrangements
should be made for clinical follow-up of the patient. Such large skin doses are most
likely to be imparted by prolonged fluoroscopically-guided interventional procedures
such as angioplasties, cardiac radiofrequency ablations, transjugular intrahepatic
portosystemic shunt (TIPS) procedures, and vascular embolizations, particularly
311
in procedures on large patients. If erythema does not appear approximately 10 to

15days after the procedure or if moist desquamation does not appear within 8weeks,
further follow-up is likely unnecessary (see chapter 20).
DICOM and IHE standards (discussed in Chapter 5) will ultimately permit
the automatic transfer of patient dose information from fluoroscopes to the PACS
or other information system. Ideally, software will become available providing
maps of doses on each patients skin. Today, it is common practice to establish
a cumulative air kerma threshold (e.g., 3 or 5 gray) and arrange for follow-up of
each patient whose procedure exceeds that threshold. For older fluoroscopes that
do not display cumulative air kerma, similar thresholds should be established for
fluoroscopy time.
SUGGESTED READING
American Association of Physicist in Medicine Radiation Protection Task Group no. 6. Managing the use of
fluoroscopy in medical institutions. AAPM Report No. 58. Madison, WI: Medical Physics, 1998.
Code of Federal Regulations 21 Food and Drugs. 21 CFR 1020.32, Office of the Federal Register National
Archives and Records Administration. April 1, 1994.
Medical Exposure of Patients in Ionizing Radiation Exposure of the Population of the United States NCRP
publication no. 160. Bethesda, MD: National Council on Radiation Protection and Measurement,
March 2009.
Proceedings of the ACR/FDA Workshop on Fluoroscopy, Strategies for Improvement in Performance, Radiation
Safety and Control. American College of Radiology, Reston, VA, October 1617, 1992.
Radiation Dose Management for Fluoroscopically-Guided Interventional Procedures, NCRP Report
No.168. Bethesda, MD: National Council on Radiation Protection and Measurement, July 2010.
Schueler BA. The AAPM/RSNA physics tutorial for residents: general overview of fluoroscopic imaging.
RadioGraphics 2000;20:11151126.
Van Lysel MS. The AAPM/RSNA physics tutorial for residents: fluoroscopy: optical coupling and the video
system. RadioGraphics 2000;20:17691786.
Wagner LK, Archer BR. Minimizing risks from fluoroscopic x-rays: bioeffects, instrumentation and examination,
4th ed. Woodlands, TX: RM Partnership, 2004.
Wang J, Blackburn TJ. The AAPM/RSNA physics tutorial for residents: x-ray image intensifiers for
fluoroscopy. RadioGraphics 2000;20:14711477.
Chapter
10
Computed Tomography
10.1 Clinical Use
From the first 80 3 80 matrix, 3-bit computed tomography (CT) images of the early
1970s, CT scanner capabilities and image quality have grown steadily over the past
40 years. Scanner rotation times have plummeted from the early 4.5-min scans to
subhalf-second scanner rotation of today, a greater than 500-fold increase in the
mechanical speed of gantry rotation. Increased gantry rotation speed combined with
an increasing number of detector arrays on the CT gantry has led to the acquisition
times per CT image going from 135 seconds per image in 1972 to greater than 200
images per second today. The reduced scan time, increased x-ray tube power, and
advanced reconstruction algorithms have led to significant advances in CT image
quality, as shown in Figure 10-1. Once called computed axial tomography (CAT)
scanners, modern CT systems can provide isotropic spatial resolution resulting in
excellent three-dimensional (3D) image data sets, including axial, coronal, and sagittal images (Fig. 10-2).
Due to the advances in image quality and reduction in acquisition time, CT
has experienced enormous growth in clinical use over the past three decades. Most
reports put the number of CT scans performed annually in the United States between
70 and 80million. The speed of CT image acquisition increased with the introduction of helical (or spiral) scanning in the late 1980s, and another large increase in
scanner speed occurred when multislice CT scanners became available in the late
1990s. Each of these advances led to pronounced increases in the utilization rates as
seen in Figure 10-3. Utilization rates (the slope of the curve shown in Figure 10-3)
increased as CT scanner acquisition speed increased, because reducing scan time
increases the number of clinical applications for CT imaging. Reduced scan time also
means that one CT scanner can image more patients per day.
10.2 CT System Designs
Gantry Geometries
Figure 10-4 shows a pediatric patient undergoing a head CT scan. For most CT
scans, a full 360-degree rotation of the CT gantry is used to collect data. The gantry
is the moving part of the scanner apparatus; however, there is a large housing that
surrounds the moving gantry to prevent mechanical injury to patients due to the
rapidly rotating hardware. In earlier scanners, electrical cables were connected from
the stationary frame to the moving frame, and the outer rim of the gantry served as
a cable take-up spool. These cables connected the stationary x-ray generator to the
rotating x-ray tube, and the signal data from the detector arrays were transferred
312
Chapter 10 Computed Tomography
313
FIGURE 10-1 A collection of CT images is shown, demonstrating the wide use of CT and the excellent
anatomical detail that CT provides.
off of the rotating gantry via cables as well. All modern CT scanners use a slip ring
connection (Fig. 10-5) between the stationary and the rotating frames, allowing the
gantry to rotate freely, untethered to the stationary frame.
The patient table is an intrinsic component of the CT scanner as well. The CT
computer controls table motion using precision motors with telemetry feedback, for
patient positioning and CT scanning. The patient table (or couch) can be removed
from the bore of the CT gantry and lowered to allow the patient to comfortably get
on the table, usually in the supine position as shown in Figure 10-6A. Under the
technologists control, the system then moves the table upward and inserts the table
FIGURE 10-2 Multiple detector array CT has led to a drastic reduction in slice thickness, and this means
that the spatial resolution of CT is nearly isotropic. Thus, in addition to the natively reconstructed axial images,
high-resolution coronal and sagittal images are routinely produced and used.
314

80
CT usage (millions)
70
60
50
40
Multi-slice CT
30
20
Helical CT
10
0
1980
1985
1990
1995
2000
2005
2010
Year
FIGURE 10-3 The excellent image quality and rapid acquisition time of CT has led to a dramatic increase in
utilization rates in recent years. Major technological advances in CT including the introduction of helical (spiral)
and multidetector array CT led to reductions in scan time, opening up CT for more clinical applications. Hence,
the use of CT in the United States has grown steadily over the past three decades. (Courtesy SE McKenney.)
with the patient into the bore of the scanner (Fig. 10-6B). A series of laser lights
provide reference to allow the patient to be centered in the bore and to adjust the
patient longitudinally. The cranial caudal axis of the patient is parallel to the z-axis
of the CT scanner (Fig. 10-7). The scanner field of view (FOV) is a circle in the x-y
FIGURE 10-4 This figure illustrates a pediatric patient in position to receive a head CT scan. Control
pads are located on either side of the gantry to allow the CT operator to position the patient. The operator
console (inset) is positioned behind a large lead glass window, providing an excellent view of the scanner
and the patient. A power injector is seen to the patients right side, hanging from the ceiling. Photo credit:
Emi Manning UC Davis Health System
315
FIGURE 10-5 A slip ring is

a component in all modern CT
scanners and allows the rotating
gantry to have electrical connections to the stationary components. Slip rings have concentric
metal bands that are connected
to a series of gliding contacts. Slip
ring technology enables helical
(spiral) CT scanning modes, and
these have led to major reductions in CT scan times.
contactors to
stationary frame
individual channels
bore
gantry
patient table
A
CT x-ray beam
centering is important
moves to position patient and

also translates during scanning
raises patient up
to center in FOV
B
FIGURE 10-6 A. The gantry is comprised of the imaging hardware, and the patient table can move up
and down and translate. The gantry is lowered to allow the patient to get on or off. B. The patient table is
raised and translated, allowing the patient to be inserted into the bore of the scanner. Laser lights are used for
patient positioning, in all three orthogonal planes. During the actual scan, the CT imaging hardware rotates
inside the gantry cover, and the table and patient translate through the bore of the gantry.
316
z-axis of a CT scanner
FIGURE 10-7 The CT image is a circle inside of the square 512 3 512 image matrix. The circular field of
view over the length of the patient forms a volume field of view that is cylindrical in shape. Most CT scanners
have a maximum circular field of view of 50 to 70 cm in diameter, but there are few restrictions in terms of
the scan length (in z).
plane but can extend considerably along the z-axis, essentially forming a cylindrical
FOV, which envelopes the patient.
In the first years of CT, the x-ray tube and detectors used a linear scanning trajectory, as illustrated in Figure 10-8. The data that were collected in this scanning
pattern correspond to attenuation measurements that are parallel to each other, and
this geometry is called a parallel beam projection. Parallel beam projection is a more
general geometrical concept, and it has utility in some nuclear and x-ray tomography geometries till today. For example, some modern scanners reorganize their
acquired data in order to use parallel ray algorithms for image reconstruction. A
projection refers to the data collected at a specific angle of interrogation of the object,
and this term is synonymous with the terms profile or view. Rays are individual
attenuation measurements that correspond to a line through the object defined at
one end by the x-ray source and at the other end by a detector. A projection is a
collection of rays.
Fan beam projection is shown in Figure 10-9. For a single detector array CT
scanner, the geometry shown in Figure 10-9 is a true fan beam geometry. The individual rays in this geometry each correspond to a line integral that spans between
the x-ray source and the x-ray detector; however, there are a lot more detectors in
the fan beam geometry compared to the parallel beam geometry. The true fan beam
geometry caveat refers to the 2D geometry shown in the figure, where there is only
minimal divergence of the x-ray beam trajectory in and out of the plane of the figure.
Figure 10-10 shows a more accurate depiction of a modern multislice CT scanner,
where the fan angle defines the fan beam geometry, but there is nonnegligible x-ray
FIGURE 10-8 In earlier CT scanners, the x-ray
tube and detector(s) translated to cover the patient,
and this geometry lead to the formation of parallel
beam geometry. Although scanners no longer use
this acquisition geometry, the concept of the parallel beam geometry remains useful in the science of
reconstruction. A projection is a collection of rays.
X-ray tube
rays
detector
parallel beam projection
317
FIGURE 10-9 A fan beam projection

refers to a fan of data that converges on
a vertex, and using rotate-rotate geometry
as shown in this figure, the apex of the fan
is the x-ray tube. The individual rays correspond to each detector measurement. The
collection of rays in this geometry is a fan
beam projection.
rays
fan beam projection
beam divergence in the orthogonal direction, which gives rise to the concept of narrow cone beam geometry. For typical 64 to 128 detector array CT scanners, the fan
angle is approximately 60 degrees, while the full cone angle is about 2.4 degrees.
Most clinical CT scanners have detector arrays that are arranged in an arc relative to the x-ray tube, as shown in Figure 10-10. This arc is efficient from several
standpointsfirstly, by mounting the detector modules on a support structure that
is aligned along a radius of curvature emanating from the x-ray source, there is very
little difference in fluence to the detectors due to the inverse square law. Very importantly, the primary x-rays strike the detector elements in a nearly normal manner, and
so there are effectively no lateral positioning errors due to x-ray beam parallax.
FIGURE 10-10 As modern scanners use more detector arrays, their
width in the z-axis dimension gives
rise to a narrow cone beam geometry, as illustrated here. Cone beam
reconstruction algorithms are used
on all modern MDCT scanners.
fan angle
cone angle
z-axis
318

FIGURE 10-11 Some commercial cone beam
CT scanners use flat panel detectors, and these
systems may be considered as true cone beam
scanners, with cone half angles approaching 10
degrees. These systems are used for image guidance on radiation therapy treatment machines,
for cone beam dental CT, and for other niche
CT applications such as breast, extremity, and
SPECT/CT systems.
z-axis
cone beam projection
Some niche CT scanners make use of flat panel detectors for the detector array,
and this is shown in Figure 10-11. This geometry represents a full cone beam geometry, where the cone angle is almost as great as the fan angle. The use of the planar flat
panel detector system leads to a 2D bank of detectors, which has no curvature, and
so correction methods are needed for the different distances from source to detector, giving rise to inverse square law and heel effectbased differences in fluence to
each detector element in the 2D array. There is also some parallax that occurs in this
geometry. The cone angle that exists in some true cone beam scanners can be appreciable; for example, using a standard 30 3 40 cm flat panel detector at a sourceto-detector distance of 90 cm, the fan angle spans about 25 degrees and the cone
angle spans about 19 degrees. Flat panel detectors are used for cone beam systems
in dental and maxillofacial CT systems, breast CT systems, and in radiation therapy
imaging applications where the cone beam CT system is mounted orthogonally to
the high-energy linear accelerator treatment beam.
Basic Concepts and Definitions

Figure 10-12 illustrates some basic geometry pertinent to CT scanners. The isocenter
is the center of rotation of the CT gantry, and in most cases (but not all), the isocenter is also the center of the reconstructed CT imagethat is, pixel (256, 256) on the
512 3 512 reconstructed CT image. The maximum FOV is defined by the physical
extent of the curved detector arrays (the fan angle). The source-to-isocenter distance
is illustrated as A in Figure 10-12, and the source-to-detector distance is labeled B.
The magnification factor from the isocenter to the detectors is then given by M 5
B/A. If the (minimum) CT slice thickness is quoted on a scanner as, for example,
0.50 mm, the actual detector width is larger by the magnification factor M. Thus, if
B < 95 cm and A < 50 cm, M 5 1.9 and the physical width of the detector arrays is
M 3 0.50 mm 5 0.95 mm. Hence by convention, most dimensions in CT are related
to the plane of the isocenter.
Figure 10-12B illustrates the rotation of the gantry around the circular FOV.
Almost all modern CT scanners have a design where the x-ray tube and detector are
attached rigidly to the rotating gantry, and this leads to a so-called third-generation
or rotate-rotate geometry. Thus, as the gantry rotates, the x-ray tube and detector
arrays stay in a rigid alignment with each other. This fixed geometry allows the use

x-ray tube
isocenter
max FOV
319
FIGURE 10-12 A. The general geometry of a CT scanner is illustrated, with the

source-to-detector distance of B and the
source-to-isocenter distance A. The magnification factor at the scanners isocenter is
M = B/A. Detector dimensions in CT are generally referred to their size as projected to the
isocenter. The maximum field of view (FOV)
is shown, although scanners routinely reconstruct smaller FOVs for heads and smaller
patients. B. The majority of modern MDCT
systems use a rotate-rotate (third generation)
geometry, where the x-ray tube and detector arrays are mounted rigidly onto the same
rotating platform and rotate in unison.
detector arrays
of an antiscatter grid in the detector array, as shown in Figure 10-13. The grid septa
are aligned with the dead space between individual detectors, in an effort to preserve
the geometrical detection efficiency of the system. As the width of the x-ray beam
has increased with multidetector array CT (to 40, 80 mm, and larger), there is greater
need for more aggressive scatter suppression. At least one manufacturer uses a 2D
grid on its large multidetector array CT system.
X-ray Tubes, Filters, and Collimation

X-ray tubes for CT scanners have much more power than tubes used for radiography
or fluoroscopy, and consequently, they are quite expensive. It is not unusual to replace
an x-ray tube every 9 to 12 months on CT systems. CT x-ray tubes have a power rating of about 5 to 7 megaJoule (MJ), whereas a standard radiographic room may use
a 0.3- to 0.5-MJ x-ray tube. There were many advancements in x-ray tube power
during the decade of the 1990s. With the advent of multidetector array CT in the
late 1990s, x-ray tube power issues abated somewhat because multislice CT scanners
320
FIGURE 10-13 Most CT systems use some

form of antiscatter technology such as the grid
pictured here. The grid septa are aligned with
the detector dead spaces to reduce impact on
detection efficiency. A one-dimensional grid is
shown here, although some manufacturers use
a two-dimensional grid. This figure shows the
concept, actual grid spacing on CT systems, is
much tighter.
one-dimensional grid
make much more efficient use of the x-ray tube output by opening up the collimation.
In changing from a 10-mm collimated beam width to a 40-mm beam width, a fourfold increase (400%) in effective x-ray beam output results (Fig. 10-14).
CT gantry rotation speeds are approaching 5 rotations per second (0.20 s rotation
time), and these high angular velocities create enormous g-forces on the components
that rotate. The x-ray tube is mounted onto the gantry such that the plane of the
anode disk is parallel to the plane of gantry rotation (Fig. 10-14A), which is necessary
to reduce gyroscopic effects that would add significant torque to the rotating anode if
mounted otherwise. Furthermore, this configuration means that the anode-cathode
axis and thus the heel effect run parallel to the z-axis of the scanner. This eliminates
heel effectinduced spectral changes along the fan angle, which is important. The
angular x-ray output from an x-ray tube can be very wide in the dimension parallel
to the anode disk but is quite limited in the anode-cathode dimension (Fig. 10-15),
and so this x-ray tube orientation is necessary given the approximately 60-degree fan
beam of current scanners.
cathode
A. tube position relative to gantry
anode
B. X-ray beam profile along z
FIGURE 10-14 A. The x-ray tube rotates on the gantry as shown in this figure. This tube orientation is
required to reduce gyroscopic effects. The anode-cathode axis runs parallel to the z-axis. B. The x-ray beam
profile in the z-dimension of the scanner includes the heel effect, since this is the direction of the anodecathode axis.
321
focal track
anode
cathode
A. wide angular range

no heel effect
(fan angle)
B. limited angular range

has heel effect
(cone angle)
FIGURE 10-15 This figure demonstrates the angular coverage of an x-ray tube. In the plane of the fan angle
(A), the x-ray beam coverage is quite wide, whereas in the direction parallel to the z-axis (B), the coverage is
limited due to the anode angle and resulting heel effect.
Most modern CT scanners use a continuous output x-ray source; the x-ray beam
is not pulsed during the scan (some exceptions are described later). The detector
array sampling time in effect becomes the acquisition interval for each CT projection
that is acquired, and the sampling dwell times typically run between 0.2 and 0.5
ms, meaning that between 1,000 and 3,000 projections are acquired per 360-degree
rotation for a -s gantry rotation period. For a typical CT system with about a 50-cm
source-to-isocenter distance, the circle that defines the x-ray tube trajectory is about
3,140 mm (2pr) in circumference. Therefore, using a 2-kHz detector sampling rate,
for example, the x-ray focal spot moves about 3,140/2,000 < 1.5 mm along the circumference of the circle per sample. While the x-ray detectors are moving in unison
with the x-ray source, the patient is stationary, and thus the 1.5-mm circumferential
displacement of the x-ray source over the time it takes to acquire one projection
can lead to motion blurring and a loss of spatial resolution. To compensate for this,
some scanners use a magnetic steering system for the electrons as they leave the
cathode and strike the anode. With clockwise rotation of the x-ray tube in the gantry
(Fig.10-16A), the electron beam striking the anode is steered counterclockwise in
synchrony with the detector acquisition interval (Fig. 10-16B), and this effectively
stops the motion of the x-ray source and helps to preserve spatial resolution. Notice
that given the 1- to 2-mm overall dimensions of an x-ray focal spot in a CT scanner
x-ray tube, steering the spot a distance of approximately 1.5 mm is realistic in consideration of the physical dimensions of the anode and cathode structures.
One CT manufacturer also uses a focal spot steering approach to provide oversampling in the z-dimension of the scan. This x-ray tube design combines high
power with the ability to stagger the position of the x-ray focal spot inside the x-ray
tube, as illustrated in Figure 10-17A. The electron beam is steered using magnetic
fields provided by steering coils, and due to the steep anode angle, modulating the
electron beam landing location on the focal track causes an apparent shift in the
source position along the z-axis of the scanner. Another manufacturer uses a different
magnetic electron beam steering approach (Figure 10-17B) to achieve a similar effect.
322

x-ray tube
anode
gantry
steered e beam
focal track
FIGURE 10-16 A. As the x-ray tube and detector rotate around the table and patient, they can travel on the
order of 1 to 2 mm per acquisition cycle, leading to potential motion artifacts. Some manufacturers compensate for this motion by steering the electron beam (B) between the cathode and the anode to rotate the focal
spot in the opposite direction, partially correcting for gantry motion.
For helical (spiral) scans, this approach leads to an oversampling in the z-dimension,
which leads to Nyquistappropriate sampling in z (see Chapter 4).
X-ray CT scanners typically run at 120 kV for generic scanning; however, more
recently, the tube voltage has been used more aggressively to optimize the trade-off
between image quality and radiation dose to the patient. The high kV combined
with the added filtration in the x-ray tube leads to a relatively hard x-ray spectrum.
The tube voltage options in CT depend on the vendor; however, 80- 100- 120- and
140-kV spectra are typical (but not exclusive) in the industry (Fig. 10-18). The high
x-ray energy used in CT is important for understanding what physical properties of
steering coils
anode
bearings
vacuum
motor
shaft
cooling oil
cathode
electron beams
shift in z-dimension
FIGURE 10-17 A. As the table advances through the gantry, the x-ray beam is rapidly shifted along z using
the x-ray design shown in this figure. The magnetic steering coils in the tube assembly shift the electron beam
trajectory, resulting in a different location for the focal spot. This design allows oversampling in the z-dimension, improving the spatial resolution in z. Using N actual detectors arrays (in z), with two focal spot positions
(about a half detector width apart), a total of 2N detector channels are formed.
323
electrostatic
collimators
dynamic
focal spot
electron beam
cathode assembly
cathode
assembly
B. anode / cathode
C. electrostatically
steered electron beam
FIGURE 10-17 (Continued) B. The alignment of the anode and cathode is shown. C. Another vendor
approach to focal spot steering uses a more conventional x-ray tube geometry, where electrostatic collimators
are used to steer the electron beam to alter its trajectory from cathode to anode. (courtesy J. Hsieh).
Fluence (mm2)
tissue are being displayed on CT images. The effective energy of the 80-kV spectrum
is about 40 keV, while that of the 140-kV spectrum is approximately 60 keV. The
range of effective x-ray energies in typical CT spectra is shown in Figure 10-19,
overlaid on the mass attenuation coefficients for soft tissue. In this region, it is seen
that Rayleigh scattering has the lowest interaction probability and the Compton
scatter interaction has the highest interaction probability. For the 120- to 140-kV
spectra, the Compton scattering interaction is 10-fold more likely than the photoelectric effect in soft tissue. For soft tissue, the CT image depicts physical properties
for which Compton scattering is most dependent on electron density. As discussed
in Chapter 3, the Compton scatter linear attenuation coefficient is proportional to
80 kV
20
100 kV
40
60
80
100
120
140 0
20
Fluence (mm2)
120 kV
20
40
60
80
100
120
140
120
140
X-ray Energy (keV)
X-ray Energy (keV)

140 kV
40
60
80
100
X-ray Energy (keV)
120
140 0
20
40
60
80
100
X-ray Energy (keV)
FIGURE 10-18 This figure illustrates typical x-ray spectra that are used by several of the CT manufacturers. Tube
voltages of 90, 130, and other slight variations are used by some vendors. The x-ray spectra in CT scanners are
hardened using significant thicknesses (5 to 10 mm Al) of added filtration, in addition to the beam shaping filters.
324

1.0E+04
Soft Tissue
1.0E+03
CT effective energy region
1.0E+02
1.0E+01
total
1.0E+00
1.0E01
Compton
1.0E02
Rayleigh
1.0E03
photoelectric
1.0E04
0
20
40
60
80
Energy (keV)
100
120
140
FIGURE 10-19 The mass attenuation coefficient for soft tissue is shown as a function of x-ray beam energy,
and the various interaction cross sections are shown. For the energy region of CT (,40 to 60 keV effective
energy), it is seen that the Compton scatter interaction cross section is much higher than the Rayleigh or
photoelectric cross sections. Hence, for soft tissue, the CT image is essentially a map of the parameter that
primarily impacts the Compton scatter cross sectionelectron density.
mCompton r N
Z

A
[10-1]
where mCompton is the linear attenuation coefficient for the Compton scattering interaction,
r is the mass density of the tissue in a voxel, N is Avogadros number (6.023 31023),
Z is the atomic number, and A is the atomic mass. The primary constituents of soft tissue are carbon, hydrogen, oxygen, and nitrogen. It is interesting to note that in all of
these elements except hydrogen, the ratio (Z/A) is for commonly occurring isotopes.
For hydrogen, the Z/A ratio is 1. While this would imply that hydrogenous tissues such
as adipose tissue would have a higher mCompton, in reality, the range of fluctuation in
hydrogen content in tissues is small and the lower density of adipose (r < 0.94 g/cm3)
relative to soft tissue (r < 1) tends to dominate Equation 10-1 when it comes to the
difference between soft tissue and adipose tissues. Consequently, adipose tissue appears
darker (has a lower mCompton) than soft tissues such as liver or other organ parenchyma.
Figure 10-19 illustrates the attenuation characteristics of soft tissue, but the
relative attenuation coefficients for soft tissue, bone, and iodine are shown in
Figure 10-20. Also shown is the range of effective energies for most CT spectra.
Because of the higher proportion of photoelectric interactions (relative to Compton)
in bone and iodine due to their higher Z, the overall mass attenuation coefficients
are higher. The mass attenuation coefficient is defined at unit density, and the higher
density of bone (r < 2 to 3) relative to soft tissue further increases the linear attenuation coefficient m (the linear attenuation coefficient m is the product of the mass
attenuation coefficient [m/r] and density, r). Hence, it is clear from Figure 10-20 that
bone and pure iodine have larger m values than soft tissue, water, or adipose.
The gray scale in CT images is a quantitatively meaningful value, unlike any
other clinical imaging modality used today. The gray scale values in CT are called
Hounsfield units (HUs), in honor of Sir Godfrey Hounsfield who was one of the principal innovators of CT technology in its early days. The HU is defined as

HU ( x, y, z ) 1000
(m( x, y, z ) - mw )
mw
[10-2]

effective energies
80 kV
140 kV
iodine
bone
325
FIGURE 10-20 This figure

shows the mass attenuation coefficient for bone, tissue, and iodine
as a function of x-ray energy. The
attenuation coefficient is much
greater for bone and greater
still for 100% iodine. Typically,
tissues have much lower concentrations of iodine. This figure
illustrates the basis for contrast in
CT between bone-, tissue-, and
contrast-enhanced CT.
tissue
where m(x,y,z) is the average linear attenuation coefficient for a volume element
(voxel) of tissue in the patient at location (x,y,z) (Fig. 10-21). HU(x,y,z) represents the gray scale CT images in the same (x,y,z) spatial coordinates, and mw is
the linear attenuation coefficient of water for the x-ray spectrum used. Note that
when m(x,y,z) corresponds to a voxel of water, then the numerator in Equation
10-2 is zero, and HU 5 0. Because m(x,y,z) is essentially zero for air, the ratio on
the right in Equation 10-2 becomes 21, and therefore HU 5 21,000 for air. The
HU range is defined only at these two pointswater and air. In practice, adipose
tissues typically range from HU 5 280 to 230, and most organ parenchyma runs
from approximately HU 5 130 to 1220. Tissue with iodinated contrast and bone
can run much higher, to a maximum of HU 5 3,095 for most 12-bit CT scanners.
Because urine in the urinary bladder is predominately water, it should be close to
HU 5 0. At all x-ray tube voltages, HUwater 0 and HUair 21,000; however, due
to energy-dependent attenuation, the HU values of adipose and soft tissues will
shift slightly with different tube voltages. This shift should be most apparent for
bone and iodine contrast.
The CT reconstruction process converts the raw acquired data from the CT scanner to a series of CT images, usually reconstructed as a series of contiguous axial
images. While each of the individual CT images is 2D, in the patients body that image
corresponds to a 3D cross section of tissue as shown in Figure 10-21A. Hence, when
referring to the location in the body, the term volume element (voxel) is used. Axial and
voxel
Dz
Dz
A. in the patient
FIGURE 10-21 A. The dimensions of the inplane (Dx, Dy) and the slice thickness in the z-axis
(Dz) are combined to form the volume element or
voxel. In modern MDCT scanners, the dimensions
of the voxel are approximately equal on thin-slice
reconstructions, where Dx = Dy < Dz.
326
pixel
y
x
pixel
square pixels on axial images
rectangular pixels on coronal images
B. in the images
FIGURE 10-21 (Continued) B. Despite the three-dimensional nature of most CT scan data, the individual
images are fundamentally two dimensional. Like all digital images, each picture element is called a pixel. The
pixel on an axial image (x,y) is square; however, the coronal image (x, z) and sagittal image (y, z) usually have
rectangular pixel which can be resampled along z such that Dx = Dy = Dz.
without beam shaping filter
x-ray fluence
approximately circular
patient cross section
b
profile along fan angle
A
FIGURE 10-22 A. Most body areas scanned by a CT scanner are circular in cross section, and this gives rise
to an uncorrected x-ray fluence at the detector that is high at the edges and low in the center, as seen in the
profile. A beam shaping filter, also called a bow tie filter, is used on all whole body CT scanners and is located
in the x-ray tube assembly. The shape of the filter is designed to attenuate more toward the periphery of the
field, which tends to make the signal levels at the detector more homogeneous.
327
with beam shaping filter
x-ray fluence
periphery dose in patient is reduced

with no loss of image quality
profile along fan angle
me
diu
lar
ge
sm
all
C
FIGURE 10-22 (Continued) B. This creates the opportunity to reduce dose to the patient with no loss of
image quality. C. A photograph of a beam shaping filter assembly from a commercial CT scanner is shown.
This assembly has three different bow tie filters, as labeled. (Courtesy J. Hsieh.)
coronal CT images are illustrated in Figure 10-21B with one picture element (pixel)
indicated. The 2D pixels in each image correspond to 3D voxels in the patient.
The vast majority of CT scans are either of the head or of the torso, and the torso
is typically broken up into the chest, abdomen, and pelvis. These exams represent
over three fourths of all CT procedures in a typical institution. All of these body
parts are either round or approximately round in shape. Figure 10-22A shows a
typical scan geometry where a circular body part is being scanned. The shape of the
attenuated (primary) x-ray beam, which reaches the detectors after attenuation in
the patient, is shown in this figure as well, and this shape is due to the attenuation
of the patient when no corrective measures are taken. Figure 10-22B shows how the
x-ray beam striking the detector array is made more uniform when a beam shaping
filter is used. The beam shaping filter, often called a bow tie filter due to its shape,
reduces the intensity of the incident x-ray beam in the periphery of the x-ray field
where the attenuation path through the patient is generally thinner. This tends to
equalize or flatten the x-ray fluence that reaches the detector array. An assembly from
328
a commercial whole body CT scanner employing three different bow tie filters is
illustrated in Figure 10-22C.
The bow tie filter has consequences both on patient dose and on image quality.
Figure 10-23A shows a central ray (C) and a peripheral ray (P) with no bow tie filter
in place. The dose at the entrance points for these two rays is only slightly different
due to the slight differences in distance from the x-ray source (i.e., the inverse square
law). The dose at the exit points in the central and peripheral rays can be markedly
different, however, due to the much longer attenuation path in the central ray compared to that of the peripheral ray. Thus, the dose at the exit point of the peripheral
ray is much greater. When these factors are applied and the dose distribution from
an entire rotation of the x-ray tube around the patient is considered, when no bow tie
filter is used, increased radiation dose at the periphery of the patient results, as shown
in Figure 10-23B. An ideal bow tie filter equalizes dose to the patient perfectly, as
entrance points
exit points
A. Central (c ) and peripheral (p) rays
high dose
high dose
B. no bow tie
C. ideal bow tie
D. Small bow tie
FIGURE 10-23 A. The peripheral ray (marked P) passes through a shorter path of tissue to reach the exit
point than the central ray (C) on a circular object, and thus the dose at the exit point for ray P is greater than for
ray C. When the source is rotated 360 degrees around the object, this consideration leads to higher dose levels
at the periphery (B) for larger cylindrical objects when no beam shaping filter is used. C. With an ideal bow tie
filter, the attenuation thicknesses of the object are exactly compensated for, producing a nearly homogeneous
dose distribution in the patient. D. If a bow tie filter designed for a small cylinder is used for a larger cylinder
(such as using a head bow tie for a body scan), the dose will be too concentrated toward the center of the fan
angle, and higher doses will result centrally.
329
shown in Figure 10-23C. If the bow tie filter has too much peripheral beam attenuation, then it will concentrate the radiation toward the center of the patient and lead
to higher central doses as in Figure 10-23D.
Since CT images are reconstructed from many projection data acquired around
the patient, the image noise at a given pixel in the CT image is the consequence of
the noise from all of the projection data which intersect that pixel. The noise variance
(s2) at a point in the CT image results from the propagation of noise variance from
the individual projections (p1, p2, , pN). A simplified mathematical description of
the noise propagation is given by

2
s CTimage
s 2p1 s 2p 2 s 2p 3 s 2pN
[10-3]
For the peripheral beams (such as P in Fig. 10-23A), in the absence of a bow tie
filter, the x-ray fluence striking the detectors is high, and thus the noise variance is
low. However, because the noise in the CT image results from all the views, some of
the projections through any point in the CT image will be from central rays (ray C in
Fig. 10-23A), and these rays have lower x-ray fluence and higher statistical variance.
Thus, the noise in the reconstructed CT image will be dominated by these higher
noise central rays. Consequently, the higher dose that is delivered to the periphery of
the patient is essentially wasted and does not contribute to better image quality (i.e.,
lower noise). Hence, the bow tie filter is a very useful tool because it reduces patient
dose with no loss of image quality. The bow tie filter is used in all commercial whole
body CT scanners and in all CT acquisition scenarios.
The standard adult head is about 17 cm in diameter, whereas the standard adult
torso ranges from 24 up to 45 cm or greater, with an average diameter ranging
between 26 and 30 cm. Because of these large differences, all commercial CT scanners make use of a minimum of two bow tie filtersa head and a body bow tie. The
head bow tie filter is also used in pediatric body imaging on most scanners.
Detector Arrays
All modern multidetector array CT scanners use indirect (scintillating) solid-state
detectors. Intensifying screens used for radiography are composed of rare earth crystals (such as Gd2O2S) packed into a binder, which holds the screen together. To
improve the detection efficiency of the scintillator material for CT imaging, the scintillation crystals (Gd2S2O and other materials as well) are sintered to increase physical
density and light output. The process of sintering involves heating up the phosphor
crystals to just below their melting point for relatively long periods of time (h), and
there are many other details to the process. In the end, densification occurs, and the
initial scintillating powder is converted into a high-density ceramic. The ceramic
phosphor is then scored with a saw or laser to create a number of individual detector
elements in a detector module, for example, 64 3 64 detector elements. An opaque
filler is pressed into the space between detector elements to reduce optical cross talk
between detectors. The entire fabrication includes a photodiode in contact with the
ceramic detector (Fig. 10-24).
The detector module sits on top of a larger stack of electronic modules, which
provide power to the detector array and receive the electronic signals from each
photodiode (Fig. 10-25). The electronics module has gain channels for each
detector in the module and also contains the analog-to-digital converter, which
converts the amplified electronic signal to a digital number. The detector array
modules are mounted onto an aluminum frame on the mechanical gantry assembly.
330
perspective view
x-ray photon
opaque filler
ceramic scintillator
light photons
photodiode
amplified
electronic signal
substrate
side view
FIGURE 10-24 The detectors in modern CT are fabricated as detector modules (perspective view). A side
view of the modern CT detector shows individual detector elements coupled to photodiodes that are layered
on substrate electronics. Spaces between detector elements are scored out creating a small void, which is filled
with an optical filler material to reduce detector cross talk. CT detectors are much more expensive than those
used in radiography and make use of sintered phosphors resulting in high-density ceramic detector arrays. The
increased density improves x-ray detection efficiency.
The aluminum frame forms an approximate radius of curvature from the position of
the x-ray source. Each detector module can be removed and exchanged on the gantry
for rapid repair, when necessary.
Figure 10-26 illustrates cross sections through the x-ray beam and detector
array for three different models of CT scannersa single array CT (for comparison),
detector array
electronics
FIGURE 10-25 A. The detector arrays are mounted on electronics module, which includes power supplies for the
amplifiers, amplification circuits for the detector module, and analog-to-digital converter systems. B. A photograph
of a detector module with electronics from a commercial CT scanner is shown. (Photograph courtesy J. Hsieh.)
collimator
331
collimator
collimator
isocenter
beam width
beam width
beam width
slice width
(at isocenter)
A. single array CT
B. multiple detector CT
C. cone beam CT
FIGURE 10-26 A. A cross section (see inset figure showing orientation) of a single array CT scanner is
shown. Adjustment of slice thickness in single detector CT involved the relatively simple manipulation of
the x-ray beam width using the collimator. B. Multiple detector array CT (MDCT) systems result in a divorce
between slice thickness and x-ray beam width. The detector array width (T) determines the minimum CT slice
width, while the overall x-ray beam thickness (.nT) is determined by the collimator. C. Cone beam CT systems
are essentially MDCT systems that have a very large number of detector arrays. While typical state-of-the-art
MDCT systems may use 64 to 128 detector channels, some manufacturers have expanded this to 256 to 320
element detector arrays (T = 0.5 mm), which give rise to a collimated beam width of 160 mm.
a small multiple detector array CT (MDCT), and a larger MDCT system. In a single
detector array scanner, the detectors were 13 to 16 mm wide, and the x-ray beam collimation was used to determine the slice thickness. Collimator settings ranged from
1 mm (thin slice) to 10 mm (thick slice) or larger. The single detector array system
integrated the x-ray signal over the entire width of each detector, and so changing
slice thickness was solely a function of the collimation. Thin slices could be acquired,
but a greater number of CT acquisitions were required to interrogate a given length of
the patients anatomy. For example, to scan 20 cm in the abdomen with contiguous
images, a total of 100 CT images need to be acquired for 2-mm slice thicknesses (t 5
2 mm). For a typical 1-s scan time, this would require 100 s of actual scan time (longer in practice due to breathing gaps). That was generally too long to be practical,
and thus 7- or 10-mm slices would typically be acquired, requiring 28 scans (28 s)
or 20 scans (20 s), respectively.
With MDCT, the slice thickness and x-ray beam width are decoupled. The slice
thickness is determined by the detector configuration, and the x-ray beam width is
determined by the collimator. In the standard vernacular of CT, the detector thickness is T (measured at isocenter) and the number of detectors is n. For one manufacturers 64 slice (n) scanner, T 5 0.625 mm and so the collimated x-ray beam width
is nT 5 40 mm. For contiguous images, the 20-cm section of abdomen discussed in
the paragraph above could be acquired in just 5 scans (5 3 40 mm 5 200 mm), and
a total of 256 CT images, each 0.625 mm thick, would be acquired. So compared
to the single slice CT system, the scan time in this example went from 100 to 5 s,
332
40 5 mm slices
320 0.625 mm slices
FIGURE 10-27 Slice thickness plays an important role in regard to the relationship between radiation dose
and CT image noise. Although very thin CT images can be produced by all modern MDCT scanners, the user
needs to balance this advantage of spatial resolution in the z-dimension with the clinical necessity for low-dose
CT imaging. One CT acquisition can produce thin-section relatively noisy images (0.625) that have high z-axis
resolution, and thick-section lower noise images can also be produced for primary interpretation.
and the slice thickness went from 2 to 0.625 mm. These capabilities underscore the
advances that MDCT offersfaster scans and thinner slices.
Dose and Image Noise Consequences of CT Section Thickness

Take an example where a top of the line MDCT (64 slice or larger) scans 200 mm of
anatomy with nT 5 40 mm. Five gantry rotations (pitch 5 1, as discussed later) are
required*, and with a 0.5-s scan, this takes 2.5 s. A total of 320 very thin (0.625 mm)
CT scans can be reconstructed, delivering excellent spatial resolution along the z-axis
of the patient. Compare this to combining the acquired projection data from 8 contiguous detectors (8 3 0.625 mm 5 5 mm), and reconstructing 5-mm thick CT images,
which would lead to 40 images covering the same 200-mm section of anatomy (Fig.
10-27). For a fixed technique (e.g., 120 kV, 300 mA, and pitch 5 1.0 for a helical
scan), the radiation dose to the patient will be the same. Indeed, the two image data
sets described above were reconstructed from the same physical CT acquisition, so
of course the dose was the same. However, the thinner T 5 0.625 mm images are
eight times thinner than the T 5 5 mm images, and consequently each thin CT image
makes use of eight times fewer detected x-ray photons. The T 5 0.625 mm images
provide more spatial resolution in the z-axis. However, the use of an eightfold reduction in photons means that they are a factor of 8 5 2.8 times noisier. This can be
seen in Figure 10-27. One can acquire thin T 5 0.625 mm images with the same
noise levels as the 5-mm images, by increasing the dose by a factor of 8. For example,
at 120 kV as before, increase the mA to 600 (23), increase the scan time from 0.5 to
1.0 s (23), and go to a pitch of 0.5 (23). However, given heightened concerns about
the radiation dose levels in CT, using thicker CT slices for interpretation is a good
way to reduce noise while keeping dose levels low.
The message of the above observations is this: Even though MDCT scanners
allow a vast number of thin images to be reconstructed, there is a trade-off in terms of
image noise, patient dose, and z-axis resolution, and that trade-off needs to balance
patient dose with image quality (resolution and noise).
Actually six rotations are required in helical mode on some systems, as discussed
later.

detector array
a
point in patient
333
FIGURE 10-28 In MDCT scanners

operating in helical (spiral) acquisition mode with the pitch near 1.0,
essentially all of the detector arrays
contribute to the raw data acquisition for each CT image. Because of
this, the beam profile for all detector
arrays needs to be approximately the
same, otherwise, artifacts will result.
This means that the x-ray beam penumbra needs to be located outside
of the active detector array, resulting in some dose inefficiencies with
MDCT.
table motion
detector array
Overbeaming and Geometrical Efficiency

For helical (spiral) scanning on a MDCT system, a given voxel in the patient is
reconstructed from the data acquired by most or all of the detector arrays as the
patients body translates through the gantry. Figure 10-28 shows how the rightmost
detector array (b) acquires the x-ray projection of a given point in the patient, but as
the table advances through the gantry and as the gantry rotates, a different detector
array (a) acquires the x-ray projection through this point in the patient. If the z-axis
beam width or shape differed significantly from detector array to detector array in
the system, then the projection data used to reconstruct each CT image would differ around 360 degrees and artifacts would occur. The artifact would be most pronounced when there is an abrupt change in attenuation in the z-dimension such as
at the end of a bone. The x-ray beam has relatively constant intensity throughout its
center, but the consequences of the finite focal spot and the highly magnified collimator give rise to a penumbra at the edges of the beam (Fig. 10-29). Because the
beam shape is so different in the penumbra region compared to the center of the
beam, the penumbra region is to be avoided, and hence the penumbra is positioned
outside of the active detector arraysthose x-rays essentially strike lead shielding on
the sides of the detector assembly or inactive detectors. This is called overbeaming.
Note that for a single detector array scanner, this is not necessary. The consequence
of overbeaming is that for MDCT scanners, the geometrical efficiency is less than the
100% geometrical efficiency of a single detector array scanner. In the early days of
MDCT scanners when 4 or 8 detector arrays were used, the geometrical efficiency
was quite low (,70%), and consequently there was a considerable radiation dose
penalty (,30%) that resulted. As more detector arrays were used and as the beam
width in the z-dimension consequently increased, the penumbra region of the beam
became a smaller fraction of the overall x-ray beam flux, and the geometrical efficiency increased as a result. With state-of-the-art MDCT scanners (64 detector
channels), the geometrical efficiency is greater than 95%.
Selection of CT Slice Width

As mentioned previously (Fig. 10-26), for single detector array CT, the slice width
was selected by adjusting the collimator width. With MDCT, the collimator adjusts
334
beam
detected
detected
beam
penumbra
z-axis beam efficiency (%)
105
100
95
90
85
80
75
70
65
60
data are approximated
55
50
0
detector arrays
A. The shape of the x-ray beam and position
of x-ray detector arrays are illustrated
16
32 48 64 80 96 112 128
Number of Detector Arrays
B. The relative geometric dose efficiency as a

function of detector arrays is shown
FIGURE 10-29 A. The z-axis dose profile of a typical MDCT scanner is illustrated and the influence of heel
effect is seen. The beam profile slopes off dramatically in the penumbra, and therefore the penumbra edge
is placed outside the active detector arrays. The x-ray beam that does not strike active detectors represents a
reduction of geometrical dose efficiency. B. The dose efficiency is shown as a function of the number of the
detector arrays. In the early days of MDCT systems with only a few detector arrays (n < 4 to 16), the wasted
dose in the penumbra was a considerable fraction (up to 30%) of the total x-ray beam. As detector arrays
become wider, the fraction of the beam represented by the penumbra is reduced, giving rise to very acceptable geometrical efficiencies for MDCT systems with more numerous arrays.
the overall beam width (nT) and the slice thickness is governed by the width of the
detector arrays (T). For a state-of-the-art scanner (typically 64 detector channels),
each detector array has its own data channel and all of the detector array signals are
recorded during a CT scan. Lets use the example of a 64-slice scanner with 0.5-mmwide detector arrays (at isocenter). All of the data are recorded, and so the system can
be instructed to reconstruct 64 3 0.50 mm slices, or 32 3 1.0 mm, 163 2.0 mm,
8 3 4.0 mm, 4 3 8.0 mm, etc. Any combination of signals from adjacent detector
arrays is possible, although the actual selections on commercial scanners are limited
to standard, usable combinations. So assuming that the acquired raw data exist on
the CT scanner console, one can go back days later and re-reconstruct the CT slice
thickness to any selectable value on the scanner. However, the raw CT data only
reside on the scanners computer for a period of time until it is overwritten with
new datahow long that is depends on how large the data drive is and how many
CT scans are performed with the system in a dayin busy settings, the raw data are
available for a day or less. With these scanners, it is typical to reconstruct 5.0-mm
axial slices for radiologist interpretation, and 0.50- to 0.625-mm axial slices to be
used for reformatting to coronal or sagittal data sets.
In older MDCT scanners with 4 to 32 detector arrays, the number of data acquisition channels in many cases was less than the number of detector arrays. For example, one vendors 16-detector scanner had only 4 data channels. In this setting, the
desired slice thickness had to be selected prior to the CT scan. For the 1.25-mm
detector width, the user could acquire 4 3 1.25 mm for a total collimated beam
width of 5.0 mm. The signals from adjacent detector arrays can be combined, and so
by binning together signals from two adjacent detector arrays, the scanner could
acquire 4 3 2.5 mm data (10.0-mm collimation). By binning 3 and 4 adjacent detector signals, 4 3 3.75 (15-mm collimation) or 4 3 5.0 mm (20-mm collimation)
could also be acquired, respectively (Fig. 10-30).
335
four electronics channels

collimated x-ray beam
z-axis
1.25 mm
4 1.25 mm
4 2.50 mm
4 3.75 mm
4 5.00 mm
detector configuration
FIGURE 10-30 For the early MDCT systems, the number of electronics channels was less than the number
of detector arrays, and so a decision of CT slice thickness had to be made prior to acquisition. For a standard
16-slice CT scanner as illustrated here, with 4 channels of electronics, the detectors could be used individually
(4 3 1.25 mm data sets) or could be binned to produce thicker slices and better CT coverage (e.g., 4 3 5.0 mm
data sets).
10.3 Modes of CT Acquisition
Scanned Projection Radiograph

Once the patient is on the table and the table is moved into the gantry bore, the
technologist performs a preliminary scan called the CT radiograph. This image is
also called the scout view, topogram, scanogram, or localizer; however, some of these
terms are copyrighted to specific vendors. The CT radiograph is acquired with the
CT x-ray tube and detector arrays stationary, the patient is translated through the
gantry, and a digital radiographic image is generated from this line-scan data. CT systems can scan anterior-posterior (AP), posterior-anterior (PA), or lateral. It is routine
to use one CT radiograph for patient alignment; however, some institutions use the
lateral localizer as well to assure patient centering. The PA CT radiograph is preferred
over the AP to reduce breast dose in women and girls.
Using the CT radiograph, the CT technologist uses the software on the CT console to set up the CT scan geometry (Fig. 10-31). The process for doing this is specific to each vendors CT scanner; however, a common theme is to place guidelines to
define each end of the CT scan. It is at this point in the scanning procedure that all
the CT scan parameters are set, usually using preset protocols. For a basic CT scan,
these parameters include kV, mA, gantry rotation time(s), type of scan (helical or
axial), direction of scan, pitch, detector configuration, reconstruction kernels(s), mA
modulation parameters, if used, and so on. Complexity increases for more advanced
scanning protocols.
Basic Axial/Sequential Acquisition

The axial (also called sequential) CT scan is the basic step-and-shoot mode of a CT
scanner (Fig. 10-32). The gantry rotates at typical rotation speeds of 0.5 s or so, but
336
FIGURE 10-31 The CT radiograph, known by a number

of other names, is typically acquired first and is used by the
technologist to set up the extent of the CT exam. Bounding
lines for an abdominal pelvis CT scan are shown.
CT radiograph
the x-ray tube is not turned on all the time. The table is stationary during the axial
data acquisition sequences. The system acquires 360 degrees of projection data with
the x-ray tube activated, the tube is deactivated, the table is moved with the x-ray
beam off, another scan is acquired, and so on. This process is repeated until the entire
beam off, move table
patient
on table
gantry
Axial or Sequential CT Acquisition

FIGURE 10-32 Axial (also called sequential) CT scanning is the basic step-and-shoot mode of the CT
s cannerthe x-ray beam is not on while the patient is being translated between acquisition cycles. For
MDCT systems, the table advance between each scan is designed to allow the production of contiguous CT
images in the z-axis direction. Thus, for a scanner with 64 (n) detector arrays with width T = 0.625 mm, the
table is moved a distance of about nT between axial acquisitionsabout 40 mm in this example.
337
FIGURE 10-33 For axial (or

sequential) CT imaging with
contiguous spacing, (A) while
the table increment results in
contiguous images, because
the x-ray beam is slightly wider
than the beam width (nT), there
is overlap of the x-ray beam
(B) between locations that
increases the radiation dose to
the patient in these regions.
A. contiguous axial imaging
B. dose overlap between scans
anatomical area is covered. Because the table and patient are stationary during an axial
scan, the x-ray tube trajectory defines a perfect circle around the patient. Due to the
tables start/stop sequence, axial CT requires more acquisition time than helical scanning (described below). With the advent of MDCT, it is common to acquire contiguous
CT images during an axial acquisition. This implies that between each axial acquisition, the table moves a distance D, which is essentially equal to the width of the detector array at isocenter (nT). This results in a series of images in the CT volume data set,
which are contiguous and evenly spaced along the z-direction. In practice, the x-ray
beam width is slightly wider than the distance nT, and so the series of axial scans
results in some x-ray beam overlap between the axial acquisitions (Fig. 10-33).
Helical (Spiral) Acquisition

With helical (also called spiral) scanning, the table moves at a constant speed while
the gantry rotates around the patient. This geometry results in the x-ray source forming a helix around the patient, as shown in Figure 10-34. The advantage of helical
table feed per 360

gantry rotation
width of active
detector arrays
Helical or Spiral CT Acquisition

FIGURE 10-34 With helical (or spiral) CT scanning, the x-ray tube has a helical trajectory around the patients
body, as illustrated. There is an important relationship between the width of the active detector arrays used
during the scan and the table feed per rotation of the gantrythis relationship is defined at the pitch.
338
scanning is speedby eliminating the start/stop motion of the table as in axial CT,
there are no inertial constraints to the procedure. Similar to the threads on a screw,
the pitch describes the relative advancement of the CT table per rotation of the gantry.
The pitch of the helical scan is defined as

pitch
Ftable

nT
[10-4]
where Ftable is the table feed distance per 360-degree rotation of the gantry and nT is
the nominal collimated beam width. For example, for a 40-mm (nT) detector width
in z and a 0.5-s rotation time for the gantry, a pitch of 1.0 would be obtained if the
table moved 80 mm/s. For most CT scanning, the pitch can range between 0.75 and
1.5; however, some vendors give more flexibility in pitch selection than others. A
pitch of 1.0 corresponds in principle to contiguous axial CT. A pitch lower than 1.0
(Fig. 10-35B) results in overscanning the patient and hence higher radiation dose to
the patient than a pitch of 1.0, all other factors being equal. A pitch greater than 1.0
represents underscanning (Fig. 10-35C), and results in lower radiation dose to the
patient. The relationship between relative dose and pitch is given by
dose
1

pitch
[10-5]
Pitch settings near 1.5 allow for faster scanning and are used for thoracic or pediatric CT scanning where speed is important. Low pitch values are used in cardiac
imaging, or when a very large patient is to be scanned and the other technique factors
(kV and mAs) are already maximized.
Because the table is moving along the z-axis during helical acquisition, there is a
slight motion artifact that reduces the spatial resolution very slightly. However, this
concern is generally balanced by the speed of the helical CT scan and the corresponding reduction in patient motion blurring that results. This is especially important in
the torso, where the beating heart and churning intestines cannot be voluntarily
stopped in even the most compliant of patients.
The acquired data at the very beginning and end of a helical scan do not have sufficient angular coverage to reconstruct artifact-free CT images. Consequently, along
the z-direction, helical scans start nT before an axial scan would start, and stop
nT after an axial scan would end. For example, for a scan where nT 5 20 mm and
there is a desired scan length of 200 mm, an axial scan would scan only the desired
A. axial / sequential
B. low pitch
helical / spiral
C. high pitch
helical / spiral
FIGURE 10-35 A. A series of axial or sequential CT scans results in a number of circular x-ray tube trajectories
around the patient, requiring the x-ray tube to be turned off between each axial scan. For helical (or spiral) scanning, the helix can be tight corresponding to a low-pitch study (B), or it can be loose corresponding to a highpitch CT study (C). Pitch values in conventional (noncardiac) CT scanning run typically between 0.75 and 1.5.

helical CT scan
helical CT scan
intermediate
beam
start beam
339
adaptive
collimation
stop beam
start beam
stop beam
dose wasted
usable data
usable data
A. no adaptive beam collimation
position 1
B. adaptive beam collimation
position 2
position 3
C. adaptive beam collimators

FIGURE 10-36 For helical (or spiral) CT scanning without adaptive beam collimation (A), half of the x-ray
beam width at both ends of the scan is not used because those data do not have sufficient angular sampling
to produce high-quality CT images, and this results in dose inefficiency. The inefficiency is worse for wider
x-ray beams and for shorter scan lengths. B. With adaptive beam collimation, collimators are used to eliminate
patient exposure at the edges of the scan, improving the dose efficiency of the helical CT acquisition. C. Adaptive beam collimators are illustrated in these photographs. The vertical rods are cam shaped, and independent
rotation provides the effect illustrated in (B). (Courtesy J. Hsieh.)
200 mm, but a helical scan would require 220 mm of acquisition to reconstruct
that same 200 mm of image data. Thus, in this example, 10% of the radiation dose
to the patient is essentially wasted. Newer helical systems, however, have adaptive
beam collimation (various trade names exist) as shown in Figure 10-36. With this
shielding, the wasted dose at the outer edges of a helical scan is collimated out.
The amount of dose that is saved as a percentage of the overall dose increases for
scans of smaller scan length, and for CT scanners with larger collimated beam widths
(nT). Some manufacturers achieve adaptive beam collimation using an additional
340
set of motorized collimators in the x-ray tube head assembly, while others use a
cam-shaped collimator, which can rotate to achieve the same effect (Fig. 10-36C).
The speed of a CT scan using a modern CT system is quite remarkable, and this
is one of the key reasons why CT is so widely used for clinical diagnosis. An example
will illustrate this. An abdomen-pelvis CT scan requires a scan length of 48 cm on a
tall patient. Using a modern MDCT scanner with a collimated nominal beam width
of 40 mm, a pitch of 1.0, and a gantry rotation time of 0.5 s, the scan can be completed in 6 s, well within a breath-hold for most patients.
Cone Beam Acquisition

All top of the line (64 detector channels) CT systems are technically cone beam CT
systems, as the reconstruction algorithm used for converting the raw data into CT
images takes into account the cone angle. However, most conventional CT systems still
make use of only slight cone angles on the order of 2 degrees (cone half-angle). Some
vendors, however, have developed what could be called true cone beam CT scanners
where the half cone angle approaches 10 degrees. For example, one vendor has a 320detector array system with 0.5-mm detectors, so the entire z-axis coverage is 16 cm
(Fig. 10-37). There are of course benefits and tradeoffs with such a wide cone beam
design. The challenges of wide cone beam imaging include increased x-ray scatter and
increased cone beam artifacts. The benefit of this design, however, is that whole organ
imaging can be achieved without table motionso for CT angiography (CTA) or perfusion studies, the anatomical coverage is sufficient for most organs (head, kidneys, heart,
etc.), and so whole organ perfusion imaging with high temporal resolution is possible.
Cardiac CT
The challenges of imaging the heart have to do with temporal resolution. With a typical heart rate of 60 beats per minute, one cardiac cycle is about 1 s. To freeze cardiac
motion, an image acquisition time window of 100 ms or less is required. With that as
Full Cone Beam CT Acquisition

FIGURE 10-37 For some CT scanners such as the 320-detector system, which has a 160-mm-wide x-ray
beam at the isocenter of the scanner, cone beam acquisition strategies are used for whole organ imaging. This
is essentially an axial CT scan that is capable of acquiring the entire data set along z, without table translation,
owing to the very large detector array and cone angle. Other full cone beam systems used in radiation oncology and dental CT use flat panel detectors that extend up to 300 mm in width.
341
a goal, scientists in the 1980s developed an electron beam CT (EBCT) scanner, which
could image the heart with a scan time of about 50 ms. This scanner had a series of
semicircular tungsten rings below the patient. An electron beam accelerated at approximately 120 kV was magnetically steered, such that it rotated around the tungsten rings
very rapidly, forming a dynamically moving x-ray source for the EBCT scanner. An
annular array of detectors was mounted above the patient. For true CT reconstruction,
data only need to be acquired 180 degrees plus the fan angle. For the heart, the fan
angle is small, and so the requirements are essentially 180 to 190 degrees for a properly
centered patient. This unconventional EBCT scanner set the standards for high-speed
CT imaging in an era when contemporary CT technology required 3- to 5-s CT scans
per image. However, modern third-generation CT systems have developed to the point
where cardiac imaging can be performed on more conventional CT hardware.
To image the entire heart using most current rotate-rotate (third-generation) CT
scanners, cardiac gating techniques are required. With cardiac gating methods, the
image data for CT reconstruction are acquired over several cardiac cycles. Gating
methods are made more challenging if the cardiac cycle changes from cycle to cycle,
and thus patients with regular heat beats are the ideal candidates for cardiac CT.
To steady the heart rate of patients with irregular heart beat, beta-blockers can be
administered prior to imaging, and this has been found to produce high-quality cardiac CT images in many settings.
Figure 10-38A shows the general imaging geometry for cardiac imaging. When
third-generation cardiac scanners were introduced, it was common to use retrospective gating techniques (Fig. 10-38B). With this approach, the heart was imaged continuously with the rapidly rotating CT scanner, and the electrocardiogram (ECG)
data were recorded in synchrony with the CT data acquisition. Because the heart
is longer in the z-dimension than what many MDCT scanners can image (nT), the
procedure for imaging the heart requires data acquisition at several table positions,
so that the entire length of the heart is imaged. As long as the cardiac cycle is not in
complete synchrony with the scanners gantry rotation speed, after several cardiac
cycles, enough data would be acquired over different phases of the cardiac cycle to
reconstruct the heart. With retrospective reconstruction, projection data are acquired
FIGURE 10-38 A. Cardiac CT systems use multiple rotations of data
acquisition over time to acquire the
data necessary for reconstructing highquality images of the heart.
A. Cardiac CT
342
1 sec
X-ray on
B. retrospective cardiac gating
X-ray on
1 sec
100 ms
C. prospective cardiac gating
FIGURE 10-38 (Continued) B. Early cardiac CT acquisition modes used retrospective gating, where the CT
scanner acquires data while it also records the electrocardiogram (ECG) information. This allowed the acquired
projection data to be selected at various time points in the cardiac cycle (gated), and these projection data
are then combined and used to reconstruct cardiac CT imageswith retrospective gating, CT images could
be produced throughout the cardiac cycle, resulting in the generation of a 3D movie of the beating heart.
However, this acquisition mode results in high dose to the patient and there is only occasional need for cardiac
imaging throughout the entire cardiac cycle. C. In prospective cardiac gating modes, the ECG is used to trigger
the x-ray tube such that projection data are only acquired during the most quiescent part of the cardiac cycle,
end-diastole. This mode reduces dose to the patient considerably and produces one cardiac CT image that is
useful for assessing coronary artery stenosis and other anatomical anomalies in the heart.
over the entire cardiac cycle. Therefore, CT images can be synthesized using the
ECG gating information over the entire cardiac cycle, and a rendered beating heart
image can be produced. In clinical practice, however, most of the interest in cardiac
CT imaging is in identifying blockages in the coronary arteries, as these represent
the greatest risk to the patient. For the evaluation of coronary stenoses, a temporal
depiction of the heart is not necessary and only the end-diastole images are necessary.
While the dynamic information achieved from retrospective cardiac imaging is
visually dramatic, it represents a high-dose procedure.
Most cardiac CT systems have advanced, so that prospective gating techniques
can be used (Fig 10-38C). With prospective gating, the ECG triggers the x-ray tube
such that projection data are acquired only during the end-diastolic period of the cardiac cyclethis is the point in the cardiac cycle where the heart is the most stationary. The benefit of prospective cardiac CT is that the radiation dose to the patient is
significantly less than with retrospective gating. The downside of prospective gating
is that a full temporal record of the beating heart is not acquired, only one snapshot
during end-diastole, but as mentioned previously, this image is usually sufficient for
evaluation of coronary artery disease. Another downside of prospective gating is that
343
if the patients heart beat is irregular during the scan, the acquisition may lead to an
incomplete CT data set, which may not be recoverable. An alternative to a full prospective mode (Fig. 10-38C) is to use higher mA during the end-diastole periods as
shown in Figure 10-38C, but to use a lower mA during the rest of the cardiac cycle,
so that images (albeit of higher noise) could be reconstructed even when an irregular
cardiac cycle is encountered during the scan.
One CT vendor has taken a more aggressive approach toward building a cardiac CT scanner, by building a CT system that has two entire imaging chains on it
(Fig.10-39). There is a large FOV imaging chain comprising an x-ray tube (Tube A) and
a full fan beam gantry, and a second imaging chain with another x-ray tube (Tube B)
and a smaller detector array is mounted almost orthogonal to the first system. The
B-imaging chain has a smaller FOV because of space limitations on the gantry. Furthermore, the FOV required for imaging the relatively small heart does not have to be
as large as a standard FOV for imaging the body. The reason why this system is well
suited for cardiac imaging is described in the next paragraph.
True CT reconstruction requires only 180 degrees plus the fan angle, and as mentioned previously, due to the small dimensions of the heart, the fan angle is small as
well. The full 360-degree gantry rotation of the dual source CT (DSCT) system is 0.33
s, and so a 180-degree rotation requires 3 0.33 s 5 166 ms. With two redundant
x-ray systems configured approximately 90 degrees with respect to each other, a quarter of a rotation of the gantry results in a full 180-degree acquisition data set when
the projection data from both imaging chains are combined. The 90-degree rotation
of the gantry requires just 3 330 ms 5 83 ms. Hence, with the speed of gantry
rotation combined with the two redundant x-ray imaging systems, the DSCT scanner
can acquire data fast enough during end-diastole to freeze the motion of the heart and
provide an excellent diagnostic tool for the diagnosis of coronary artery stenosis.
It is noted that cardiac CT requires only an intravenous injection of contrast agent,
while the competing technology coronary angiography requires intra-arterial contrast
injection, a far more invasive procedure. So for diagnostic cardiac procedures, CT
techniques offer the patient and their physician a less invasive diagnostic procedure.
However, for cardiac interventional procedures involving balloon angioplasty with or
without the placement of coronary stents, coronary angiography is still required.
tube A
FOV B
FOV A
tube B
dual source CT
FIGURE 10-39 The dual source CT system was developed specifically for cardiac
imaging. This system has two full x-ray imaging systemstwo x-ray tubes and two sets
of detector arrays; however, due to space
issues, the B imaging chain has a smaller
detector array resulting in a smaller field
of view, but still well tailored to the imaging requirements for the heart. Due to the
approximately 90-degree juxtaposition of the
two imaging chains, a 90-degree rotation of
the gantry results in the acquisition of projection data over 180 degrees, which is sufficient for true CT reconstruction at the center
of the field of view. With a 330-ms period for
360-degree gantry rotation, the 90-degree
rotation requires 330/4 = 83 ms. Thus, it is
a short enough time period to acquire a full
projection data for CT reconstruction within
a fraction of the typical cardiac cycle.
344
Dual Energy CT
In his original notes on CT, Hounsfield predicted the use of CT for dual energy
decomposition methods to separate density from the elemental composition of materials. The initial experimental work on dual energy x-ray imaging was performed on
a CT scanner in the mid-1970s, as this was the only platform at that time that could
conveniently produce digital images. Dual energy CT is similar conceptually to dual
energy projection imaging discussed in Chapter 7.
The concept of using CT for dual energy imaging has been known in the research
arena for decades, but it is likely that the dual source CT scanner described above
for cardiac imaging launched the most recent clinical interest in dual energy CT. The
DSCT scanner was designed primarily for cardiac imaging; however, the availability
of two complete x-ray systems on the same gantry clearly lends itself to the acquisition of a low and high kV data set during the same CT acquisition. On single source
CT systems, other manufacturers have developed the ability to rapidly switch the kV
to achieve dual energy acquisition in a single acquisition. However, it is more difficult
to switch the tube current (mA) with high temporal frequency since the tube current is determined by the x-ray tube filament temperature, which cannot be modulated at kilohertz rates due to the thermal inertia of the filament. This is an issue
because optimization studies of dual energy CT imaging demonstrate that the low kV
(80 kV) needs higher output levels than the high kV (140 kV) setting. However, at
the same mA, the x-ray tube is far more efficient at x-ray production at high energies.
Averaged across four major CT manufacturers, the output (air kerma) at the isocenter
of the scanner at the same mAs is about 3.3 times higher at the highest kV setting
(135 to 140 kV) compared to the lowest kV setting (80 to 90 kV). To overcome this,
single tube CT scanners increase the temporal pulse width at the lower kV setting to
increase the relative signal levels at the lower tube potential.
There are trade-offs for dual energy CT between the dual source and single source
approaches. The dual source system has a smaller Tube B FOV, which limits the FOV
of the dual energy reconstruction. Also, because both x-ray systems are operational
simultaneously on a DSCT system, scatter from one imaging chain impacts the other
and this requires correction techniques, which ultimately result in more noise in the
data. On the other hand, DSCT systems have two distinct x-ray imaging chains and
so additional added filtration can be added to the higher kV source, which increases
energy separation and improves the energy subtraction.
There are two approaches to reconstructing dual energy CT imagesthe raw
dual energy image data sets for each projection angle can be subjected to logarithmic
weighted subtraction techniques, similar to those used in projection radiography, and
then the dual energy subtracted projection data can be used to reconstruct dual energy
CT images. Alternatively, the low kV and high kV data sets can be reconstructed separately and subtraction techniques or other digital image manipulation can be applied
directly to the reconstructed CT images. In clinical practice, the latter technique is most
widely used. Note that the HU values in CT are already linear with respect to the linear
attenuation coefficient, and so no additional logarithms are necessary. Ultimately, the
dual energy image is a linear combination of the low- and high-energy CT images. Figure 10-40 illustrates the mapping of HU values between the 80- and 140-kV images.
CT Angiography
CT has higher contrast resolution than traditional angiography, and consequently
requires only venous access as opposed to more invasive arterial access for
345
aluminum
hydrogen
bone
calcium
Figure 10-40B
water
muscle
polyethylene
fat
oxygen
PMMA
teflon
nitrogen
carbon
B
FIGURE 10-40 This figure shows the Hounsfield units for a number of elements, biological compounds, and
tissues at unit density. The 80 kV HU values are plotted on the horizontal axis, while the 140 kV HU values are
plotted on the vertical axis. A. A plot covering large HU values is shown. It is noted that higher Z materials are
below the line of identity. The Zs of hydrogen, aluminum, and calcium are 1, 13, and 20, respectively. The HU
of calcium assumes 10% density. B. This plot zooms in on HU values in a range closer to water. Water has an
effective Z of 7.42, and materials with a lower effective Z than water are above the line of identity, while materials with a higher effective Z are below the line of identify. The Zs of carbon, nitrogen, and oxygen are 6, 7, and
8, respectively. Zeff of Teflon (CF2) is 8.4, polyethylene is 5.4, PMMA is 6.5, and muscle is approximately 7.6.
uoroscopy-based angiography. CTA is really just contrast-enhanced CT where the

fl
images have been processed to highlight the vascular anatomy using image processing techniques (Fig. 10-41). In general, maximum intensity projection procedures
produce excellent CTA images. Some image rendering techniques use pseudocolor
to add realism to the images.
CT Perfusion
CT perfusion imaging is used to evaluate vascular perfusion and other physiological
parameters related to blood flow to a specific organ. While most often used to evaluate stroke or vasospasm in the brain, CT perfusion is also used for abdominal organ
imaging, especially in the oncologic setting. CT perfusion is a high radiation dose
346
FIGURE 10-41 CT angiography (CTA) is

used with increasing frequency instead of
fluoroscopy-based angiography. For diagnostic procedures, the venous access for
contrast agent injection is far less invasive
than penetrating the (high-pressure) arterial system for iodine injection. CTA images
of the arteries in the leg and the head are
shown, with false color rendering.
studythe CT scanner acquires images of an organ repeatedly in real time to quantify

the flow of iodine-based contrast agent through that organ. Typical head CT perfusion
protocols call for the acquisition of about 40 CT images, each using a 1-s acquisition,
resulting in about 40 s of scanning in the same area of tissue (Fig.10-42A). CT perfusion takes advantage of the wide collimated beam width (20 to 160 mm) of top of the
line scanners (64 data channels) to image a volume of tissue in the same region of
the body repeatedly. The CT scan starts prior to the intravenous injection of contrast,
and hence the scanner produces a number of CT images prior to the arrival of the contrast bolus to the organ of interest. This is necessary because the mathematical models
(Patlak, etc.) that are used to compute the parametric maps require an input function
(HU versus time), typically the artery supplying blood to the organ of interest. From
this data set of temporal CT scans, and with physician demarcation of the location
of the input function, parametric maps are computed with high resolution. In addition
to the raw gray scale CT images, most commercial perfusion software packages compute a number of different parametric 3D volume data sets of pseudocolored images,
showing for example blood volume, tissue perfusion, and time to peak enhancement
(Fig. 10-42B). Perfusion CT is an example of where the well-appreciated anatomical
capabilities of CT are combined with the high temporal resolution capabilities of the
modality to produce physiologically meaningful, quantitative, functional images.
Shuttle Mode CT
Shuttle mode allows a CT scanner to repeatedly image a volume of tissue that is wider
than the detector array (nT). The table rocks back and forth a prescribed distance,
during the temporal acquisition procedure (Fig. 10-43). This mode allows CT perfusion analysis to be performed over a larger section of tissue. The temporal resolution
347
actual beam width

30 50 CT scans at
the same location
perfusion
A. perfusion imaging
time to peak
enhancement
blood volume
B. computed parametric images
FIGURE 10-42 CT perfusion is an extension of CT angiography, where a relatively long time sequence of
repeated CT images is acquired for physiologic assessment of various organs (A). Typically on the order of 40
full CT data sets are acquired for brain perfusion studies, requiring about 40 seconds. From these data and
some user input, (B) physiological metrics such as vascular perfusion, time to peak enhancement, and blood
volume can be computed using software designed for this purpose.
is lower in shuttle mode, but adequate physiological information can in general be

derived. Due to the change in direction of the table, shuttle mode results in image
data being acquired during table acceleration and deceleration, which implies that
the reconstruction algorithm needs to accommodate for the changing pitch.
Dose Modulation
Some body regions in the patient are rounder than other regions. An elliptical cross section of a patient is shown in Figure 10-44A. With fixed x-ray tube current and hence
actual beam width
table motion
coverage increased
using table motion
shuttle mode
FIGURE 10-43 Shuttle mode imaging is a hybrid

between CT perfusion and helical CT imagingthe
table shuttles back and forth over a relatively small
(e.g., 100 to 200 mm) predefined region in the patient,
allowing temporal imaging and hence CT perfusion
assessment to be made over a wider length of anatomy
than the width (nT) of an MDCT field of view.
348

a
A. Tube rotation around oval patient
mA
time
B. mA modulation
FIGURE 10-44 Tube current modulation is illustrated. A. As the x-ray tube and detector array rotate around
the patient, the x-ray path length is determined either by attenuation measurements or from the CT radiograph, or both. The tube current (mA) is modulated to optimize the use of radiation in the patient. The mA is
increased when the beam is projected through thicker regions, and it is decreased as the x-ray beam traverses
thinner projections through the patient. B. For an elliptical shaped body part such as in the pelvis, the mA
modulation scheme is approximately sinusoidal as the gantry rotates around the patient.
constant x-ray flux, the detectors will receive more primary x-rays when the gantry
is in position a compared to gantry position b, due to the increased attenuation path
in position b. However, as we have already seen in the context of the bow tie filter,
the noise variance in the projection data propagates into the reconstructed CT image
in a manner that is more optimal when the projection data variance levels are nearly
matched (Equation 10-3). Hence, the higher signal levels (higher dose and lower signal
variance) in position a are essentially wasted when combined with those in position
b during CT image reconstruction. When the mA is modulated, the mA is reduced in
position a compared to position b as shown in Figure 10-44B. When mA modulation is
performed, lower radiation dose levels can be used to achieve the same image quality.
The mA modulation scheme shown in Figure 10-44 occurs as the x-ray gantry
rotates once around the patient. However, as the patient translates through the gantry
over greater distances in z, the shape and size of the patients body can change appreciably. Therefore, modulation also can be applied along the z-axis of the patient. The
average mAs values are shown as a function of position along a patients torso in Figure 10-45. The angular modulation as a function of gantry angle is quasi-sinusoidal
in elliptical body parts and constant in round body regions, and these fluctuations
occur over relatively shorter distances as highlighted in Figure 10-45. The impact of
mA modulation is illustrated on CT images in Figure 10-46, where the average mAs
per CT image are indicated.
The role of mA modulation in CT is analogous to that of automatic exposure control in radiography and automatic exposure rate control in fluoroscopy. mA modulation is implemented differently for different CT vendors. One scanner manufacturer
uses the attenuation data measured by the detectors near the central ray in the scanner to adjust the target mA for 180 degrees later in the scan. Manufacturers also use
the attenuation data measured during the acquisition of the CT radiograph to plan
the mA modulation scheme for each patient. Figures 10-44 and 10-45 demonstrate
the relative mA values used, but mA modulation technology is also used to set the
absolute mA levels (and hence the patients radiation dose) for the scan.
current me (mAs)
angular mAs
modulaon
overall impact
of mA modulaon
349
FIGURE 10-45 Tube current modulation

occurs as the gantry rotates around the patient to
address the different x-ray path lengths between
the AP and the lateral projections, but it also is
active along the z-dimension of the patient. As
the scanner images the patients thorax, abdomen, and pelvis, the tube current is modulated
to optimize the image quality through the entire
extent (along z) of the patient scan. The overall
mAs per slice along the z-axis of the patient is
shown in this figure, and the oscillatory nature
of the angular mAs modulation is also shown.
(Courtesy SE McKenney.)
relave posion
Different scanner manufacturers employ very different concepts for the selection
of the overall dose levels in CT, and this can and does lead to confusion in the clinical environment. The details are beyond the scope of this book; however, a general
description is in order here. One approach is to set the actual noise level in the reconstructed CT image, and this value is used to guide the overall dose levels (by adjusting the mA levels used). With this approach, when lower noise levels are selected,
the mA values and resulting radiation doses are higher. Maximum and minimum mA
values can be set as a precaution. A different approach to mA modulation is to set a
68 mAs
105 mAs
94 mAs
122 mAs
FIGURE 10-46 The overall effect of tube current modulation is to use different radiation levels for regions
in the body with different overall attenuation, and most tube current modulation schemes result in a more
homogeneous distribution of CT noise in the patient images along the z-axis. Here, the mAs values used for
the various body regions are indicated.
350
nominal mA level that the system uses as guidance for the overall dose settings, but
in this situation, higher nominal mA values result in higher mA settings and higher
radiation dose to the patient. When used correctly, mA modulation techniques produce excellent CT images at near-optimal radiation dose levels. Therefore, it is an
important feature of modern CT scanners that should be used, especially in body
imaging. However, when the responsibility of adjusting something as important as
the radiation dose in CT is turned over to the CT computer, it is important that the
operators (technologists, radiologists, medical physicists) are well educated in the
use of the mA modulation technology on each CT scanner they use.
CT Protocols
A typical CT scanner may have 100 to 300 different preset protocols loaded on it.
The CT operator usually starts a patient scan procedure by selecting the CT protocol
appropriate to the scan for that patient. The CT protocol encompasses a wide variety
of parameters, which can be set, including CT acquisition parameters such as kV, mA,
rotation time, slice thickness, and pitch, but also the reconstruction kernel(s) used, the
timing, volume, and injection rate of the contrast injection scheme (if used), the default
window/level settings for the reconstructed CT images, and what computer to send
the images to when they are ready. The protocol is the starting point in the scanning
procedure, and once the patient is on the table and the CT scan is about to commence,
it is the role of the experienced CT technologist to modify the protocol in some cases.
For example, for very large patients, the mAs (or maximum mAs for dose modulation
schemes) need to be increased to produce a diagnostic quality CT examination.
10.4 CT Reconstruction
Preprocessing
There are a number of preprocessing procedures that are applied to the actual acquired
projection data prior to CT image reconstruction. The details of these steps are proprietary to each vendor; however, some general observations can be made. During
routine calibration of the CT scanner, the influence of the bow tie filter is characterized by performing air scans. The air scans also characterize differences in individual
detector response, which may be due to differences in photodiode or amplifier gain.
The measured projection data for a given CT scan to be reconstructed are normalized
by the calibration scans, and this procedure corrects for previously identified inhomogeneities in the field. In some scanners, a small fraction of the detector elements
may be dead, and these are routinely identified using software for this purpose.
A dead pixel correction algorithm is applied, which replaces dead pixel data with
interpolated data from surrounding pixels. Scatter correction algorithms generally
need to be applied before the logarithm of the data is applied. Adaptive noise filtration methods can be applied; for example, the scanner can invoke algorithms, which
identify regions in the projection data that correspond to low signal areas, and these
areas have high noise and after backprojection will correspond to locations of high
noise on the CT images. To reduce the impact of noise, some algorithms identify
these low signal areas and then apply smoothing or other data processing steps to
reduce the noise in these areas.
After preprocessing, the projection data undergo logarithmic transformation and
normalization, in order to correct for the exponential attenuation characteristics of
351
x-ray interaction. Each projection measurement through the patient corresponds to

a discrete measurement of Ij at detector element j, where
I j g j I o e - ( 1t 2t 3t ....nt )
[10-6]
The signal in a reference detector Ir located outside the FOV of the patient is also
measured during the CT scan:

I r gr I o
[10-7]
It is known during the calibration air scans that the ratio of the gains between a
detector j and the reference detector r is given by b 5 gr/gj. This ratio can not only
compensate for differences in actual gain between detectors, but also correct for
differences in bow tie filter thickness, distance from the x-ray source (i.e., inverse
square law), etc.
I
P j ln r t (m1 m2 m3 ..... mn )
b I j
[10-8]
From Equation 10-8, we see that the projection measurements, Pj, after preprocessing and logarithmic conversion, correspond to linear sums of the linear attenuation
coefficients through the patient and along the path of the ray, and this is illustrated
in Figure 10-47 as well.
Simple Backprojection
To start the discussion as to how the CT image is computed from the projection data
sets, lets examine a simplified but realistic example of the problem (Fig. 10-48). As
discussed above, the logarithm converts the exponential nature of x-ray absorption
into a linear problem, and we will therefore deal with this example as a linear problem. The 3 3 3 grid of numbers in the central boxes in Figure 10-48 are the data
of interestthese numbers represent the object, and the image of the object. Four
projections through this patient are shown, ad. It can be seen that the projected
value through each line passing through the patient is the linear sum of the voxel
values that each ray intersects. Creating the projections from the central data is called
forward projection, and that is what the CT scanner hardware does physically. Mathematical forward projection is used in iterative reconstruction techniques and many
other reconstruction processes as well. The problem that is posed by Figure 10-48
X-ray source
m1
m2
m3
m4
.
mN
detector
Ir
Ij
FIGURE 10-47 The reference detector is positioned to detect

the output of the CT scanner with no patient anatomy in the
measurement, resulting in Ir. A large number of ray measurements Ij are made for detector elements j = 1, N. The ray
measurements sample a number of individual linear attenuation (m) values along the beam path, as illustrated. While x-ray
attenuation is an exponential process, logarithmic processing
as described in the text converts the measured data to linear
projections of m.
352
3
d
2
4
10
10
a
FIGURE 10-48 The 3 3 3 grid in the center of this figure represents the actual attenuation values (scaled
to integer values for demonstration), and the projection data (a, b, c, and d) show how the attenuation values
are summed in this very simplified example. The ad projections are the acquired data in a CT scan, and the
reconstruction challenge is to determine the values in the nine elements in the 3 3 3 grid. For this simple
example, the process is similar to Sudoku. The reader is challenged to use the projection values (only) to find
a solution for the values in the gold rectangles.
is,if you erase the nine values in the 3 3 3 box, can you figure out what they are
from the projection data? This looks very much like a Sudoku problem at this point,
and given the simplicity of the 3 3 3 problem, it can be solved by trial and error.
However, in a CT scanner, the matrix is typically 512 3 512, and the task of reconstructing the values in the image from the projection values is more formidable.
We saw that to produce the projection values, knowing the contents of the image,
the straightforward process of forward projection is used. The reverse problem, computing the image matrix from the projection values, can be solved (almost) by backprojection. With backprojection, the measured projection values are simply smeared
back into the image matrix to compute the backprojected image. Figure 10-49 shows
an example of forward projection of a simple circular object at the center of the field,
and therefore the projections at each angle are identical in shape. The projection
of a circle is a parabola, and that is seen in Fig 10-49A. If that same (uncorrected)
projection shape was used to backproject the CT image from all projection angles
(Fig. 10-49B), the resulting image would have a characteristic blurring as shown in
the isometric plot in Figure 10-49C. The take-home message is this: Using simple
backprojection as described above, the reconstructed image has a characteristic 1/r
blurring that results from the geometry of backprojection. To correct for this, a mathematical filtering operation is required, and that leads to the discussion of filtered
backprojection in the next section.
Filtered Backprojection
In the case of simple backprojection, the process of summing projections from a large
number of angles around 360 degrees results in the 1/r blur function as seen graphically in Figure 10-49C. The mathematical function 1/r is illustrated in Figure10-50A.
measured projection
353
backprojection with
uncorrected projection
C. Reconstructed image
A. Measured projection
B. Backprojection reconstruction
FIGURE 10-49 For actual CT data sets, the challenge is to reconstruct a total of 206,000 cells (p 3 2562). In
this figure, a simple circle at the center of the image is imaged, producing the measured projection in A. If
these same measured projection data are used in backprojection (B), the image will experience a characteristic
1/r blurring as illustrated in C.
It is possible to correct for the impact of the 1/r function using image processing
procedures. The mathematical operation of convolution describes such a procedure,
and convolution is discussed in more detail in Chapter 4, and in Appendix F. When
undoing an effect caused by convolution, deconvolution is used. Deconvolution is
mathematically identical to convolution, except that the deconvolution kernel is (by
definition) designed to undo a specific effect. The convolution process is defined
mathematically as

p ( x )
x '-
p ( x ) h( x - x ) dx p ( x ) h( x )
[10-9]
11.0
10.0
9.0
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0
10.08.0 6.0 4.0 2.0 0.0 2.0 4.0 6.0 8.0 10.0
Filter Amplitude
Amplitude (A.u.)
where p(x) is each measured projection, h(x) is the deconvolution kernel, and p(x)
is the corrected projection value, which is used in filtered backprojection. The righthand side of Equation 10-9 shows the short hand notion for convolution, where
is the mathematical symbol for convolution.
Position (mm)
A. 1 / r blur function
30
25
20
15
10
5
0
5
10
15
2.0 1.5 1.0 0.5
0.0
0.5
1.0
1.5
2.0
Position (mm)
B. deconvolution kernel
FIGURE 10-50 A.The 1/r blurring function is illustrated. The blurring caused by the geometry of backprojection can be corrected by deconvolving the measured projection data prior to backprojection, resulting in the
process of filtered backprojection. B. The discrete implementation of the deconvolution kernel is illustrated.
354
The deconvolution kernel that is designed to undo the 1/r blurring shown in
igure 10-50A is shown in Figure 10-50B. The oscillatory nature of the deconvolution
F
kernel in Figure 10-50B is a result of the discrete nature of the operation, as applied
in a computer. As mentioned in Chapter 4, when a convolution kernel has negative
values such as that shown in Figure 10-50B, in general it sharpens an image. The
kernel in Figure 10-50B is a 1D function, and it is used to deconvolve the measured
1D projection values prior to backprojection, as illustrated in Figure 10-51. Here it is
seen that when all of the measured projections of the object are deconvolved with the
appropriate kernel, subsequent backprojection results in a faithful representation of
the object. An actual example of this is shown for a test object in Figure10-52. The
convolution operation described here is a form of mathematical filtering, and this is
why the procedure is called filtered backprojection. Convolution backprojection as
described in Equation 10-9 can be considered as a specific implementation of filtered
backprojection.
Fourier-Based Reconstruction
Convolution as described in Equation 10-9 can be performed quicker using properties of the Fourier transform. Specifically, Equation 10-9 can be recast as
p '( x ) FT -1 {FT [ p ( x ) FT [h( x )]}
[10-10]
where FT[] refers to the forward Fourier transform operation and FT-1[] is the inverse
Fourier transform. Equation 10-10 is mathematically equivalent to Equation 10-9;
however, the Fourier transform approach can be performed faster in a computer.
Therefore, the Fourier approach is used in commercial scanners more often than the
convolution approach. Notice that the Fourier transform of the convolution kernel
h(x) is performed in Equation 10-10; however, because h(x) does not change for a
given CT reconstruction filter, its Fourier transform is precomputed. Because of this,
it is very common in CT to discuss FT{h(x)}more so than discussing h(x) itself.
measured projection
backprojection with
filtered projection
C. Reconstructed image
A. Measured projection
B. Backprojection reconstruction
FIGURE 10-51 This figure shows the same object and projection data (A) as with Figure 10-49A; however,
here the measured projection data are deconvolved with the kernel illustrated in Figure 10-50B. The resulting
filtered projection data (B) are used for backprojection, and this results in an image that reflects the properties
of the original object (C).
reconstructed image
filtered back projection
simple back projection
sinogram
355
FIGURE 10-52 The properties of simple backprojection (top row) and filtered backprojection are illustrated,
(bottom row). The sinogramthe plot of the raw data showing rays horizontally and angles verticallyis
shown on the left; the sinogram on the bottom has been filtered (horizontally) and the high-pass characteristics of the filtering process are evident. Backprojection of the filtered sinogram data results in an image that
closely matches the characteristics of the object that was scanned, while the upper reconstructed image shows
significant blurring.
Figure 10-53 shows the Fourier transform of the deconvolution kernel illustrated in
Figure 10-50B, and this is the ramp filter labeled in Figure 10-53. The incredibly
simple shape of the frequency domain filter FT{h(x)} is also perhaps why it is so
widely discussed in the CT reconstruction setting.
Why is the ramp filter the filter that we start with in filtered backprojection?
Figure 10-50A shows the 1/r blurring effect that results from simple backprojection,
and it turns out that this is a 1/f effect in the frequency domain. So, if you have an
image that has a 1/f dependency that you want to correct for, the correction process
would involve multiplying the image by a function that has an f dependency, which
is described mathematically as

H( f ) f
[10-11]
where a is simply a scaling factor. The correction process results in f 3 1/f, which
eliminates any frequency dependency. It should be recognized that Equation 10-11
defines the ramp filter seen in Figure 10-53, with some very minor caveats that are
beyond the scope of this text.
As mentioned above, the ramp function H(f) shown in Figure 10-53 is the starting
point in designing the shape of the filter in filtered backprojection. It would work well
in the case of an image where there was very little statistical noise in the projection
image data sets (the measured data during CT reconstruction). However, because we
356
ramp filter
ramp filter with roll-off
Frequency (relative to Nyquist)

FIGURE 10-53 The kernel shown in Figure 10-50B is the convolution kernel that can be used in the spatial
domain to correct the measured projection values prior to backprojection. In practice, the convolution process is
implemented using Fourier-based mathematics, as described in the text. The Fourier transform of the deconvolution kernel in Figure 10-50B is shown in this figure as the ramp filter. While the ramp filter restores the blurring
that occurs during backprojection, it amplifies high frequencies and can exacerbate quantum noise in the image.
Hence, most clinical CT systems use a number of different reconstruction filters that include some roll-off,
which reduces the noise amplification effect of the ramp filter at the higher spatial frequencies. One filter is illustrated here, but there are a large number of reconstruction filters that are available on commercial CT scanners.
seek to reduce the radiation dose to patients in all x-ray imaging procedures including
CT, the dose levels are set intentionally low, and as a result, the measured data sets
are in fact noisy. Notice the shape of the ramp filter in Figure 10-53; it has high values at high spatial frequencies, and therefore significantly amplifies the high spatial
frequencies in the projection dataand this is the frequency region where quantum
noise tends to dominate. Consequently, if one were to use only the ramp filter in filtered backprojection, the reconstructed CT image would in most cases be unacceptably noisy. Therefore, it is customary to apply a function that rolls off the amplitude
of the reconstruction filter at high spatial frequencies, as shown in Figure 10-53.
The roll-off reduces the high frequency component of the reconstructed image, and
tends to abate the noise as a result. The specific roll-off function varies for the specific
application in CT. In the mathematical reconstruction literature, the names of specific roll-off filters include Shepp-Logan, Hamming, Cosine, etc. Some commercial
CT manufacturers give these mathematical filters application-specific names such as
soft tissue, bone, or standard filters. Other companies use numerical designations for
their filters, such as H47 or H50 for head filters and B40 or B70 for body filters.
Cone Beam Reconstruction

Cone beam reconstruction algorithms are similar to standard fan beam reconstruction
algorithms, but they keep track of the beam divergence in the z-direction (the cone
angle direction, as illustrated in Figures 10-10 and 10-11). Backprojection occurs
as described above; however, the algorithm computes the backprojection angles in
both the fan and the cone angles. The basic cone beam reconstruction process, often
referred to as the Feldkampt algorithm or FDK, reconstructs the entire volume data
set (a series of CT images of a given thickness) simultaneously. Cone beam reconstruction violates mathematical requirements for sampling Fourier space, and for extreme
cone beam geometries used in flat panel detectors (as in Fig. 10-11) or other very
wide cone beam CT systems, this can lead to cone beam artifacts (discussed later).
357
Iterative Reconstruction in CT
Filtered backprojection was the primary method for reconstructing CT images for
many decades; however, advancements in algorithm design coupled with fast computer hardware have led to the use of clinically useful iterative reconstruction algorithms for CT images. Iterative reconstruction is numerically intensive and has been
used clinically for smaller data sets as with nuclear medicine applications such as single photon emission computed tomography (SPECT) for many years. However, by
comparison, CT scanners acquire far more raw data than do SPECT systems, and the
matrix size is much larger (512 3 512) versus (64 3 64) for most SPECT systems. It
is observed that a 5122 matrix has a factor of 64 more pixels than a 642 matrix.
The fundamental problem in CT image reconstruction is shown in Figure 10-54,
where projection data sets are shown with a CT image. The projection data are what
is known, and the CT image is the unknown and therefore needs to be reconstructed.
The projection data sets are used to reconstruct the CT image as illustrated in Figure
10-54, but if the CT image (or an approximation of it) is known, then the projection
data can be computed using forward projection. Note that in Figure 10-54 only three
projection data sets are illustrated, but in clinical CT, a large number of projection
data are acquired (1,000 to 4,000 projections).
Iterative reconstruction techniques go through a series of iterations in the process
of CT reconstruction, as illustrated in Figure 10-55. The first image may start as a
guess, a constant image, or as a filtered backprojection image. The iteration process
then goes through a series of updates (iterations), which drive the CT image toward a
final, accurate depiction of the object that was scanned. Most iterative algorithms use
similar fundamental logicthe current estimate of the image is used to generate forward projections, which are then compared to the actual measured projections (during
acquisition) as shown in Figure 10-55. The difference between the forward projection
and the measured projection is the error matrix, which is computed for all projections
measured around 360 degrees. Each specific iterative algorithm (there are many) uses
the error matrix to update the next iteration of the image, with the intent of reducing
the error matrix in the subsequent iteration. There is a balance between reconstructing
FIGURE 10-54 This figure shows an object and three projections through the object, illustrating a small subset of
the measured projection data acquired during CT scanning. These projection data are in principle physically measured
using the detector hardware on the scanner, but if a guesstimate of the image data is known, the projections can
also be computed (as forward projections) using relatively simple mathematics (sines and cosines) in a computer.
358

iteration
iteration
iteration
forward projection
measured projection
FIGURE 10-55 The logic behind iterative CT image reconstruction is illustrated in this figure. The reconstruction process begins with an initial estimate of what the object may look like. This guess is used to
compute forward projection data sets, which are then compared with the measured projection data sets. The
difference between the mathematical projections and the physical measurements creates an error matrix for
each projection. Each type of iterative algorithm for reconstruction uses the error matrix and then updates
the next iteration of the image. The process continues for a number of iterations until the error matrix is minimized, and the CT image at that point should be a good estimate of the actual object that was scanned.
the anatomical information and the noise in the image. The many different iterative
algorithms use the error matrix to update the next iteration in a different manner. After
a certain number of iterations, the estimated image becomes an excellent depiction of
the object that was scanned. It is recognized that iterative algorithms can model many
of the physical parameters that filtered backprojection cannot, such as the x-ray spectrum, the blurring of the focal spot, etc., and therefore, in principle, iterative algorithms
can make better use of the acquired datathey can produce images with higher signalto-noise ratio (SNR) at the same dose, or they can produce images of similar SNR at
lower doses. Figure 10-56 illustrates comparisons between images reconstructed using
filtered backprojection images and iterative methods, and the iteratively reconstructed
images show a distinct advantage. Iterative methods for reconstruction do not explicitly use the filtering methods as described above, and consequently the noise texture in
the images can be quite different than with filtered backprojection images.
All major CT manufacturers have implemented iterative reconstruction techniques on their CT scanners; however, at present, this capability is an optional feature. Because of the demonstrated benefit of potential dose reduction using iterative
reconstruction techniques, CT reconstruction methods will gradually shift toward
iterative methods in order to derive better image qualitylower noise CT images
using lower radiation doses.
10.5 Image Quality in CT
Spatial Resolution
Spatial resolution in CT results from the fundamental resolution properties of the
image acquisition, as well as the resolution characteristics of the reconstruction filter
that is used. Factors that determine the ultimate resolution include the focal spot

FBP
359
MBIR
WW
2000
WW
400
standard FBP
statistical iterative
reconstruction
model-based iterative
reconstruction
FIGURE 10-56 Iterative reconstruction in CT is made possible by algorithms that are computationally efficient and require very few iterations. The iterative technique for reconstruction can in most cases extract more
signal-to-noise ratio from the acquired data set, resulting in higher SNR reconstructed images compared to
filtered backprojection techniques. Here, two images are illustrated showing the comparison from a clinical CT
system, where the interactive reconstruction methods resulted in superior noise properties compared to filtered
backprojection of the same acquired projection data. A. Filtered backprojection (FBP) is compared with modelbased iterative reconstruction (MBIR) in this figure. B. This figure shows the same image produced by standard
FBP, statistical iterative reconstruction, and model-based iterative reconstruction. (Courtesy J. Hsieh.)
size and distribution, the detector dimensions, the magnification factor, whether or
not gantry motion is compensated for, patient motion, etc. Because CT images are
the result of an elaborate mathematical reconstruction process, their resolution is
also very dependent on the characteristics of the reconstruction algorithm and any
mathematical filters used.
Spatial Resolution in the Reconstruction (x,y) Plane

The in-plane CT image is a direct product of the CT reconstruction, and by far most
clinical CT images are acquired and then natively reconstructed in the axial plane.
Some variations in patient positioning (e.g., bent neck with head tilted forward) lead to
occasional exceptions to this. Spatial resolution has historically been measured in the
clinical environment in CT using bar patterns, as illustrated in Fig 10-57. These images
were acquired using the American College of Radiology (ACR) accreditation phantom,
which has eight different bar pattern modules around the periphery of the 20-cm
diameter phantom. The bar phantom has modules corresponding to spatial frequencies of 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, and 1.2 line pairs per mm (lp/mm). Note that in
360

ACR phantom spatial resolution section
A. soft tissue kernel
B. bone kernel
FIGURE 10-57 In the field measurements of spatial resolution in CT still rely on visual assessment of bar
phantomsin this figure, the resolution module from the American College of Radiology (ACR) CT accreditation phantom is shown, imaged with (A) a soft tissue reconstruction filter and (B) with a bone filter. The same
CT acquisition can produce different images depending on which reconstruction filters are used.
most commercial CT literature, the resolution is quoted in line pairs per cm instead of
the line pairs per mm as is typical in the rest of x-ray imaging. Here, we use lp/mm for
consistency, recognizing that lp/cm 5 10 3 lp/mmthus, 12 lp/cm 5 1.2 lp/mm.
The images shown in Figure 10-57 were generated from the same CT acquisition, and the differences in spatial resolution are the result of the reconstruction filter
used. For the soft tissue image (Figure 10-57A), three of the bar patterns can be
easily resolved corresponding to a limiting spatial resolution of 0.6 lp/mm. The CT
image reconstructed with the bone kernel (Fig. 10-57B) allows six of the modules to
be resolved, for a limiting spatial resolution of 0.9 lp/mm.
The modulation transfer function (MTF) is a more scientific measure of the spatial resolution characteristics of an imaging system, as discussed in Chapter 4. The
MTF can be measured using a wire or plane of metal foil scanned on a CT scanner,
and the MTF(f) is computed from the measured line spread function, LSF(x). Five
MTF(f) curves are shown in Figure 10-58, corresponding to a specific manufacturers
reconstruction filters, as listed in the figure. It is quite apparent from this figure that
the reconstruction kernel has a significant influence on the MTF(f) and hence on the
resolving power of the images. It is noted that the limiting resolution of each kernel
is approximately at the 10% MTF level, and the limiting spatial resolutions seen in
Figure 10-57 are corroborated by the MTF curves seen in Figure 10-58.
Spatial Resolution along the Z-axis

Historically, the spatial resolution along the z-axis of the CT data set (the slice
thickness direction) has been measured using the slice sensitivity profile (SSP
Fig. 10-59A). The SSP is defined as the shape of the system response to a point input
in the z-dimension. In standard image science vernacular, the SSP is the LSF along
the z-dimension in the CT scanner. In the ACR phantom, there are a series of high
contrast wires spaced 0.5 mm apart in the z-dimension (Fig. 10-59A). Counting the
number of visible lines and multiplying by 0.5 mm results in a field measurement
of the slice thickness. Images were reconstructed at nominal slice thicknesses of 4.8
mm (Fig. 10-59B) and 2.4 mm (Fig. 10-59C), and the ACR test suggests that these
slice thicknesses measure about 5 and 3 mm, respectively.
Fig 10-57A
361
Fig 10-57B
limiting spatial resolution
FIGURE 10-58 Scientific evaluation of the resolution of CT images relies on the assessment of the modulation transfer function (MTF), as shown in this figure. This figure shows the dramatic impact on the MTF that
various reconstruction filters have. (Courtesy K. Krzymyk.)
The typical reconstructed slice thickness in the mid-1990s was approximately

7 mm, and the SSP was routinely used in that era to characterize the slice thickness.
Today it is routine to reconstruct images with 0.5- to 0.625-mm slice thicknesses,
rivaling or even finer than the voxel dimensions in the (x,y) plane. Therefore, it is
useful to use the MTF metric to characterize the spatial resolution in the z-dimension
as illustrated in Figure 10-60. For image data sets that are reconstructed into thicker
sections (e.g., T 5 2.5 mm), the MTF(f) degrades as expected. It is noted that the
z-dimension in an axially acquired CT data set can routinely have higher spatial
resolution as compared to the (x,y) axial in-plane geometry. The reconstruction filter
ACR phantom slice thickness section
0.5 mm
SSP
A. measurement concept
B. 4.8-mm slice thickness
C. 2.4-mm slice thickness
FIGURE 10-59 The spatial resolution in the z-dimension of the CT image has been traditionally measured
by the slice sensitivity profile, SSP. The ACR phantom has a series of small wires positioned at 0.5-mm intervals along the z-axis of the phantom (A). After scanning the phantom, the number of visible wires is counted
and multiplied by 0.5 mm, to estimate slice thickness. B. A 4.8-mm nominal CT scan of the ACR phantom is
showed, and 10 wires are visible suggesting that the slice thickness is approximately 5.0 mm. C. A 2.4-mm
nominal image through the ACR phantom is illustrated, with 6 wires visible suggesting a 3.0-mm measured
slice thickness.
FIGURE 10-60 The method of estimating

CT slice thickness using the ACR phantom is
a discrete implementation of the traditional
measurement of the slice sensitivity profile. The SSP is simply the line spread function (LSF) in the z-axis of the CT scan. So,
measuring the SSP and then computing the
Fourier transform of that results in the MTF
for the z-axis of the CT scanner, as shown
in this figure. The MTF shows higher spatial
resolution along z for thinner slice images,
and the z-axis resolution degrades (the MTF
amplitude is reduced) as the reconstructed
slice thickness increases.
1.2
Detail Kernel
1.0
0.8
MTF(f)z
362
0.6
t = 0.625 mm
0.4
t = 1.25 mm
0.2
t = 2.50 mm
0.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Spatial Frequency (mm1)
affects the resolution in the (x,y) plane, but it does not in the z-dimension. For this
reason, coronal and sagittal CT images have the potential for better spatial resolution
than axial CT images.
The spatial resolution in the z-axis dimension refers to the ability to see detail in the
CT images; however, it should be recognized that the dose distribution in the z-dimension is an independent parameter. As mentioned previously, the width of the x-ray beam
in all modern multidetector array CT scanners is wider than the actual detector array, in
order to exclude the penumbra region from contributing to the images and causing artifacts. The dose profile in the z-dimension can be measured using various film products
or a computed radiography (CR) plate, as illustrated in Figure 10-61. Isometric plots
erasure by CT laser system
FIGURE 10-61 The slice sensitivity profile, its discrete implementation in the ACR phantom, and the z-axis MTF
all are measures of the z-axis image resolution. However, the overall dose profile along the z-axis can only be measured using an external device such as film or a computed radiography (CR) plate, the latter of which was used
to generate the image data shown here. Various nominal collimated slice thicknesses (as set on the CT console)
were used to scan a CT plate taken out of the cassette; the CR plate was moved along z to sequentially measure
the dose profiles for different nominal slice thicknesses. The image data from the CR plate are plotted as isometric
plots on the inset image. These plots show the dose profile along z for several different collimation thicknesses.
363
shown in the figure demonstrate the estimated width of the dose profilesfor the largest
width shown, the dose profile was about 14% wider than the active detector array.
Noise Assessment in CT
Contrast Detectability Phantom: Visual Assessment
The traditional method for quantifying contrast resolution in CT is to visually assess a
contrast detail phantom, as shown in Figure 10-62. These images were produced from
the ACR phantom. The image on the left was produced using four times the radiation
as the image on the right, and the number of visible low-contrast circles is far greater.
The test objects on this phantom are designed to have 6 HU greater density (0.6%),
than the background but they vary in diameter. The smaller circular test objects have
lower SNR and therefore are theoretically and experimentally more difficult to see.
This visual assessment of the low contrast resolution of a CT scanner is a relatively
insensitive measure of the noise characteristics of a CT scanner; however, it is currently the industry standard.
Image Noise in CT: The Standard Deviation s

The CT image of a region of the phantom that is homogeneous in composition can be
used to quantify the noise directly, using software available on virtually any CT scanner or PACS system. The standard deviation s, the direct measurement of noise, is
computed using the root mean square method:
(HU
N
- HU
i 1
[10-12]
N 1
ACR phantom low contrast detectability section
s 8
120 kV, 5 mm, 285 mAs
s 16
120 kV, 5 mm, 71 mAs
FIGURE 10-62 CT images of the low-contrast module of the ACR phantom are illustrated, for radiation dose
levels varying by about a factor of 4 (285 mAs/71 mAs < 4.0). The circles visible in the phantom correspond to
0.6% differences in contrast (6 HU), and different rows of the circles have different diameters, which result in different signal-to-noise ratios (SNRs) of the circles. It is clear that more of the circular test objects are visible in the
higher dose image on the left. The standard deviation, s, is measured using software on most CT scanners or PACS
systems. The standard deviation is seen to vary by a factor of two, for these images that used a fourfold difference
in radiation dose, illustrating the Poisson relationship that image noise is relative to the square root of the dose.
364
As expected in Figure 10-62, the noise in the right image is about twice that of the
left image since the radiation levels were reduced by a factor of 4. This underscores
the relationship in quantum limited imaging where the noise (s) is proportional to
the square root of the dosefor example, double the dose and the noise goes down
by 40%, quadruple the dose and the noise goes down by a factor of 2.
Noise Texture in CT: The Noise Power Spectrum

The MTF(f) described above characterizes how efficiently signal propagates through
an imaging system, while the noise power spectrum (NPS) describes how the noise
propagates through an imaging system. The NPS(f) is a spatial frequencydependent
function as is the MTF(f). While the standard deviation s measures the overall noise
level in a region of the image, the NPS(f) describes both the overall noise level and
the noise texturea term referring to the frequency dependence of the noise in an
image. The NPS describes how the noise at one point in the image is correlated to the
noise in the surrounding points of the image. Noise correlation occurs for a variety of
reasons, but the point spread function of the imaging system tends to be a primary
culprit in causing noise correlationand because the PSF has tails that extend
from one voxel location in the image to many other voxel locations in the image
data set. Noise in the central region of the PSF is likely to therefore propagate to surrounding areas where the PSF extends. Another way to visualize this is to realize that
the resolution cell in CT is not defined by a single voxel, but extends like a cloud to
a number of voxels in the x-, y-, and z-dimensions. CT numbers (HUs) within this
cloud have a tendency to influence each other.
Noise correlation in the (x,y) plane in CT is largely due to the filtered backprojection reconstruction kernel typically used in CT. Clearly, backprojection smears
the projection data across the image, and therefore the noise in the backprojected
data impacts a large swath of pixels in the reconstructed image. In the days of the
single detector array CT scanners, however, there was little noise correlation in the
z-dimension, because adjacent images (in z) were acquired using a completely different CT scan, separated in time and space. The multiple detector arrays of modern CT
scanners combined with helical (and cone beam) CT reconstruction techniques give
rise to physical mechanisms for noise correlation in the z-dimension (in addition to
x and y). Electronic cross talk as well as PSF effects across the detector arrays in the
z-dimension clearly can induce noise correlation in z. Because of the noise correlation, which exists in all three dimensions in CT, it is necessary to talk about the 3D
NPS, which can be denoted as NPS(fx, fy, fz). To measure the 3D NPS, a volume of data
that extends some distance in x, y, and z needs to be assessed. The NPS is measured
by evaluating areas of an image that are generated from homogeneous sections of a
phantom. In practice, the 3D NPS is determined by computing the 3D Fourier transform of data acquired from a 3D volume of interest (VOI) in the CT image data set.
The mathematical details of NPS computation are beyond the scope of this chapter;
however, they are discussed in Chapter 4 and Appendix G.
In most cases, the NPS in the (x,y) plane can be approximated to be rotational symmetric, and therefore it is possible to collapse the 2D NPS(fx, fy) data into 1D using

f xy
f x2 f y2
[10-13]
This allows the depiction of the NPS(fxy, fz), as illustrated in Figure 10-63. The family of NPS curves shown in Figure 10-63A illustrate the NPS as a function of frequency on the x-axis, where this axis corresponds to the fxy frequency as described
in Equation 10-13. There are a number of different NPS curves as a function of the
1200
0.833
0.781
0.729
0.677
0.625
0.573
0.521
0.469
0.417
0.365
0.313
0.260
0.208
0.156
0.104
0.052
0.000
1000
800
NPS
365
600
400
200
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
Frequency (1/mm)
14000
0.833
0.781
0.729
0.677
0.625
0.573
0.521
0.469
0.417
0.365
0.313
0.260
0.208
0.156
0.104
0.052
0.000
12000
NPS
10000
8000
6000
4000
2000
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
Frequency (1/mm)
FIGURE 10-63 The three-dimensional noise power spectra (NPS) are illustrated for (A) a standard body
kernel. These data were computed computing the 3D Fourier transform from volumes of interest (VOIs) in a
homogeneous region of a polyethylene phantom. The family of curves illustrates the z-axis dependency of the
NPS, with the fz values listed in the legend on the right. The figure on the right (B) illustrates similar data for
the higher bandpass (bone) reconstruction kernel. (Courtesy SE McKenney.)
z -dimension, and the labels on the right side of the figure describe (using colors)
the spatial frequency in the z-dimension. Figure 10-63A shows the 3D NPS for a CT
data set reconstructed with a relatively smooth reconstruction filter (one with a lot of
roll-off, for instance for soft tissue imaging), and Fig. 10-63B shows the 3D NPS for a
higher bandpass reconstruction kernel (for instance, for bone imaging).
While the 3D NPS provides a rich description of the noise characteristics in
a CT scanner, the use of the NPS as a practical tool in clinical CT image quality
assessment is only now emerging. The NPS is a measure of noise power, and
therefore the integral over all frequencies of the NPS(fx, fy, fz) will yield the noise
variance of the CT image 2 s2.
MTF and NPS measurements on CT images assume that the images were produced using a linear system. When filtered backprojection is used to reconstruct the
images, this results in a loose approximation for a linear system. However, when
iterative reconstruction methods are used, it is likely that the CT image data strays
significantly from the definition of a linear system. It is observed that the resolution
of a CT image when reconstructed using iterative methods differs depending on the
local noise levels, and therefore the PSF is not the same from region to region, and
the system is not stationary (Chapter 4). Therefore, the MTF and NPS may not be
valid analytical methods in this context. Other than the subjective visual observation
366
methods described above, there are no obvious candidates for replacement of the
MTF and NPS metrics when iterative reconstruction is used.
Primary Factors That Affect Spatial Resolution in CT

X-ray Tube Focal Spot Distribution
The focal spot plays a role in reducing the spatial resolution in CT, as the object is highly
magnified relative to projection radiography. Also, CT systems run at very high mA
typically, and this can increase the size of the x-ray focus (focal spot blooming).
Gantry Motion
The x-ray source and detector array are moving relative to the stationary patient,
both in the angular dimension and along the z-dimension for helical acquisition.
Methods to compensate for gantry motion such as focal spot rastering (in both directions) can reduce this source of spatial resolution reduction.
Detector Size and Sampling

The detector width and sampling influence the resolution (width) and the Nyquist
limitations (sampling) to resolution. Smaller detector dimensions and oversampling
methods can improve spatial resolution.
Reconstruction Filter
Much of the reduction in spatial resolution is done intentionally, by the selection of
a reconstruction filter with significant roll-off at high spatial frequencies, in an effort
to reduce the appearance of image noise. CT images can be reconstructed multiple
times with no dose penalty to the patient, producing both high spatial resolution and
low noise image data sets.
Primary Factors That Affect Contrast Resolution

(Noise) in CT
Technique Factors
The kV, mA, time, and pitch are fundamental determinants of the dose levels used for
CT scanning, and they fundamentally impact the noise levels in the CT images. mA
and time (and hence mAs) have a linear relationship with dose, while kV does not
have a linear relationship with noise.
Slice Thickness
Thicker images combine signals from more detected x-ray quanta and therefore are
less noisy than thinner images acquired at the same technique levels. A slice thickness of 5 mm has 40% less noise
5.0 2.5 than a 2.5-mm slice thickness.
Reconstruction Filter (Filtered Backprojection)

The choice of reconstruction filter results in a fundamental and important trade-off
between spatial resolution and image noise.
Reconstruction Method
The use of iterative CT reconstruction methods can significantly reduce image noise
in comparison to filtered backprojection reconstruction. This means that lower dose
images using iterative reconstruction techniques will be equivalent to higher dose
studies using only filtered backprojection.
367
10.6 CT Image Artifacts
Beam Hardening
Standard filtered backprojection reconstruction algorithms do not fully address the
polyenergetic nature of the x-ray spectrum used in CT, but rather treat the thicknessdependent attenuation using the notion of the effective linear attenuation coefficient.
In rays that pass through a great deal of dense tissue, such as the petrous bone illustrated in the CT image inset in Figure 10-64, the high degree of attenuation causes
the x-ray spectrum to become hardened. A hard x-ray spectrum refers to one with
higher average x-ray energies, and a soft spectrum has lower average x-ray energies.
Because the presence of dense (and higher z) structures such as bones in the x-ray
path cause the lower energies in the x-ray spectrum to be preferentially attenuated
compared to the higher x-ray energy photons, the beam undergoes a upward shift in
average x-ray energy as it passes through greater thicknesses of bone. The curve in
Figure 10-64 illustrates this, and the two x-ray spectra shown illustrate the relative
removal of the lower x-ray energy photons at greater bone thicknesses.
In CT, to reduce artifacts caused by beam hardening, the x-ray beam is prehardened
with significant thicknesses of added filtration (up to 10 mm of added aluminum, for
example). However, beam hardening artifacts can occur, especially when dense objects
such as bones or metal implants exist. Figure 10-65 illustrates beam hardening in a
woman with bilateral hip implants. The x-ray beam corresponding to the paths that
intersect both metallic hip implants are exceptionally hardened, and the angular artifact
seen in the CT image (Fig. 10-65) results from this scan geometry. CT scanners do typically use simple methods for correcting for beam hardening; however, due to the simplicity of the most basic approach, artifacts can and do still occur. It is possible to use
a two-pass reconstruction method that can yield almost complete correction for beam
hardening; however, such algorithms are not usually applied in most clinical settings.
Streak Artifacts
Streak artifacts (Fig. 10-66) occur when the attenuation levels of a region in the
patient are excessive, which can exceed the dynamic range of the detector systems
or the effective linear range of the detectors. Metallic fillings in the teeth are a common source for streak artifacts, as are most implanted devices (e.g., cardiac pace-
Average Beam Energy (keV)
76
120 kV, head bowtie
74
72
70
68
66
64
62
0
10
15
20
25
Bond thickness (mm)
30
FIGURE 10-64 The concept of

beam hardening is illustrated. As
the CT beam traverses greater
thicknesses of highly attenuating
material such as bone, the average
x-ray energy changes as the graph
demonstrates. The x-ray spectrum
undergoes attenuation, and lower
energy x-rays are preferentially
attenuated, resulting in a characteristic right shift in the spectrum,
as shown on the inset spectral diagrams. Beam hardening results in
artifacts that appear as webbing
between dense regions in the
image (inset image with arrow).
368

hip implants
angles most influenced by beam hardening
plane of CT
CT through hip
localizer view
FIGURE 10-65 An extreme example of beam hardening is illustrated, from a women with two metallic hip
implants as seen in the localizer view. The CT reconstruction through one hip shows the characteristic wedge
artifact, which corresponds to the angles defined by the two metallic hips, as illustrated in the diagram.
makers, neurostimulators) that have significant metal components. Bullets and other
high-density foreign bodies also can cause considerable streaking.
Streak artifacts are exacerbated by any motion of the high-density object; Although
the head lends itself to excellent immobilization externally, during the CT scan, the
patient will sometimes swallow or move their jaw, amplifying the streak artifacts that
tend to emanate from metallic fillings.
View Aliasing
View aliasing refers to the use of too few projection images acquired to reconstruct
high-frequency objects in the image. Figure 10-67A illustrates this effect on CT
images of the head of a mouse. Most commercial CT scanners collect enough views
such that view aliasing rarely occurs; however, for some very high frequency (sharp)
objects (such as resolution test objects, Fig. 10-67B), view aliasing can occur.
Partial Volume
Partial volume artifacts occur (Fig. 10-68) when the CT voxels are large enough to
encompass several types of tissue, such as bone and tissue or tissues from different
organs. While partial volume artifacts can still occur, the dramatic reduction in the
volume of a voxel in the past decade has substantially reduced the role that partial
volume artifacts play in the diagnosis in modern CT examinations. Partial volume
FIGURE 10-66 Streak artifacts from high-density, high atomic
number dental fillings are seen in this image. Streak artifacts are
amplified in effect if there is slight motion of the high-density
object. Patients who move their jaw or swallow during a head
CT can generate significant artifacts due to this.
Views
500
167
A. Image of mouse with different numbers of views
369
FIGURE 10-67 View aliasing is the term given

to CT reconstructions that do not have proper
view sampling to render the sharper objects in the
image. A. Images of a mouse head are shown,
which were reconstructed using different numbers of views (as labeled). Most clinical CT scanners used a fixed, large number of views (950 to
3,000) for CT image reconstruction. View aliasing
on commercial CT scanners is sometimes visible
(B) when imaging high-contrast periodic structures, such as the resolution module on the ACR
phantom.
50
B. ACR resolution section
artifacts arise essentially from reconstructing low resolution images, typically thick
slice images. In many institutions, multiple reconstructions are used to produce two
axial data sets, one thick slice and one thin slice. The thin slice images can be evaluated if partial volume effects are observed by the radiologist in reading the thicker
slice images. Another method for reducing the impact of partial volume artifacts is
to reconstruct multiple CT projections, such as the axial and coronal projections.
What appears as a partial volume artifact in one projection can usually be ruled out
by visualizing the same region in the other projection.
Cone Beam Artifacts

Cone beam acquisition strategies can lead to undersampling in the cone angle dimension, and this can cause a well-known cone beam artifact (Fig. 10-69). The Defrise
phantom, which is a stack of attenuating disks separated by low density material (Fig. 10-69A), can be used to evaluate cone beam artifacts. In a well-sampled
5-mm section thickness
1.25-mm section thickness
FIGURE 10-68 Partial volume artifacts occur when the reconstructed voxel dimension is large relative to
structures in the patient. With the advent of MDCT systems, images can be reconstructed with very small slice
thicknesses, and in most cases, this eliminates partial volume artifacts. A. A 5-mm thick CT image through the
lungs are shown, and a very small lesion is seen (circle) that has lower HU values consistent with lung cancer.
B. When the same lesion is seen with thin section (1.25 mm) reconstruction, the contrast of the lesion is not
diluted due to partial volume averaging, and the high HU values are evident, consistent with a benign diagnosis. (Courtesy R. Lamba.)
370
B. Some cone beam artifacts
A. Defrise phantom
C. pronounced cone beam artifacts
FIGURE 10-69 Cone beam artifacts are accentuated by using a phantom that has alternating stacks of
high-attenuation and low-attenuation material, as pictured in A. This concept was originally described by
Michael Defrise. In a system with good sampling along z (B), only minimal cone beam artifacts are seen. However, for a true cone beam CT scanner (C), the connections between the dense areas in the image demonstrate
the cone beam artifact, which arises due to undersampling the object in Fourier space.
e nvironment, cone beam artifacts can be kept to a minimum (Fig. 10-69B); however,
the use of large cone angles can lead to considerable artifacts (Fig. 10-69C). Cone
beam artifacts are a result of fundamental deficits in the acquired data, and the most
obvious solution for these artifacts is to acquire a more complete data set.
10.7 CT Generations
Knowledge of the development of CT is useful because it demonstrates how engineering challenges were gradually overcome as new technology became available. In
addition, it is common to refer to various acquisition geometries by their generation in CT, and so this section also provides the reader with the necessary vernacular
used in the CT field.
Translate-Rotate: First- and Second-Generation CT

The first CT scanner used clinically was the EMI Mark 1, offered in 1972. By todays standards, the scanner would be virtually unusable with an 80 3 80 array of 2.4-mm pixels,
and a 3-bit (8 shades of gray) image. But 40 years ago, CT provided physicians the
nascent ability to see 3D anatomy within the body that did not require open surgery.
This first scanner was a head scanner only. An x-ray tube was mounted in a
fixed relationship with the two detector array (the first multislice CT scanner!), and
80 parallel rays were acquired as the assembly translated across the FOV. The system
rotated 1 degree, and then the x-ray tube/detector assembly translated in the other
direction (Fig. 10-70). This scanning pattern continued until 180 degrees of data
were acquired. This rotate-translate geometry using the pencil beam is referred to as
the first-generation CT. The scan required over 4 minutes and image reconstruction
ran overnight.
The success of the first EMI scanner led to the rapid development of the secondgeneration CTwhich used the same rotate-translate motion as before; however,
more detectors were added and so the initial pencil beam geometry became a narrow fan beam. A detector array using 30 detectors meant that more data could be
acquired in a given scan (Fig.10-71) , and so this led to a shorter scan time and a
larger, 160 3 160 matrix, reconstructed image.
translate
x-ray tube
371
late
trans
rotate
detector
Translate-Rotate
First & Second Generation CT
FIGURE 10-70 First- and second-generation CT systems used a translate-rotate motion for scanning the
patient. The x-ray tube and detector assembly translate in one direction acquiring a series of rays that were parallel to each other, the system rotated 1 degree, and then the x-ray tube/detector translated back in the other
direction. This scanning procedure continued until a total of 180 degrees of projection data was acquired.
Rotate-Rotate: Third-Generation CT
It was recognized by engineers at the time that the translate-rotate geometry of the
first- and second-generation CT could never lead to rapid scanning, due to the large
inertial forces involved in stopping the translation in one direction, and then reversing direction. For faster scanning, the translational motion had to go. Hence, with
third-generation CT, the x-ray tube and detector array are mounted in a fixed position with respect to each other on a rotating gantry; however, the detector array is
now long enough to allow angular sampling over the entire FOV. The rotate-rotate
geometry of the third-generation CT (Fig. 10-72A) is still the most widely used
geometry on modern scanners. With a detector array which sweeps out a fan angle
of a, and a CT system with a source to isocenter distance of S, the diameter of the
FOV is given by
FOV 2S sin(a / 2)
[10-14]
For a practical body CT scanner, the FOV should be about 50 cm in diameter, and
with a source to isocenter distance of 50 cm, the fan angle a needs to be about 60
degrees. Most modern CT scanners have geometries that roughly (620%) adhere to
these values.
FIGURE 10-71 The first-generation CT system
had just two detectors, offset in zso this first
CT scanner was really a multiple detector system.
The second-generation CT used more detectors
arranged in the x-y plane, which produced better sampling for the CT reconstruction.
Pencil Beam
First Generation CT
Narrow Fan Beam

Second Generation CT
372
360o ring of detectors
A. Rotate-Rotate
Third Generation CT
B. Rotate-Stationary
Fourth Generation CT
FIGURE 10-72 A. Third-generation CT is characterized by a gantry with rigidly mounted x-ray system and
detector arrays, in which both systems rotate together around the patientresulting in the rotate-rotate
motion B. Fourth-generation CT systems used an entire 360-degree ring of detectors mounting in the stationary frame, with a rotating x-ray tube resulting in rotate-stationary acquisition.
Cable Spools and Inertial Limitations

Earlier third-generation CT scanners required cable spools to accommodate the
cables, which connected the rotating components to the stationary components
the largest cable was the thick high-tension cable that ran between the stationary
x-ray generator and the x-ray tube, which was on the rotating gantry, but power
to motors (fans, collimator motors, etc.) and other rotating components (detectors, computers, etc.) also had to be provided, and the signals from the detectors
had to be conveyed off of the moving gantry. These cable connections between
the rotating and stationary frame required that the gantry accelerate fast, perform the scan, and then decelerate rapidly before cable lengths were exceeded.
Much of the engineering involved cable handling systems. The cable-connected
third-generation CT scanner had no inertial limitations during the acquisition of
a single CT image (unlike first- and second-generation geometry); however, large
acceleration and deceleration cycles existed between the acquisition of each CT
image, and thus CT scan times (which involve the acquisition of many CT images)
were several minutes. The minimum scan time on scanners with cable connections was about 3 s.
The introduction of the slip ring scanner (Fig. 10-5 shows a slip ring) has eliminated the cable connections between the stationary and rotating frames. However,
because the x-ray tube is powered at electrical potentials of up to 140,000 V, the slip
ring scanner requires that the x-ray transformer be located on the rotating frame to
avoid the arcing problems that a 140-kV slip ring connection would cause. To convey sufficient power across the slip ring, very high current is required. The slip ring
allows the CT gantry to rotate freely and therefore eliminates the need to deal with
gantry inertia between scans. Allowing the table to move continuously as in helical
(spiral) scanning eliminated patient inertia issues, and the elimination of inertial barriers during the entire examination led to the 0.3- to 0.5-s gantry rotation periods of
modern CT scanners.
373
source fan
detector fan
Ij
Ir
A. Third Generation CT
I j and I r
B. Ring artifact formation
C. Fourth Generation CT
FIGURE 10-73 The rigid alignment of the x-ray source and detector array in third-generation CT (A) gives
rise to the potential for ring artifact generation when the CT detectors are not properly calibrated with respect
to each other (B). In fourth-generation CT, the alignment of the x-ray tube changes with respect to the detector array and the CT data are computed from the so-called detector fan (C). Because in fourth-generation CT
each detector serves as its own reference detection, changes in detector sensitivity are factored out in the
computation of the projection data setseliminating ring artifacts.
Rotate-Stationary: Fourth-Generation CT
The third-generation CT geometry leads to the potential for ring artifact production,
due to the fixed angular relationship between the detectors and the x-ray source on
the gantry. Fourth-generation CT scanners were designed primarily to address this
issue. A fourth-generation CT scanner has a rotating x-ray tube but the entire detector array is stationary (Fig. 10-72B); hence, these systems have a rotate-stationary
geometry. The major observation with fourth-generation scanners is that an entire
360-degree array of detectors is needed, instead of the 60-degree detector array of
a third-generation scanner. Thus, a factor of six more detector elements is required,
adding significant cost.
The rationale for the fourth-generation geometry is seen in Figure 10-73. Thirdgeneration scanners (Fig. 10-73A) use a source fan geometry where the signal from two
different detectors is used to make the two x-ray intensity measurements (IrandIj)
necessary to compute a projection value, as described in Equation 10-8. If one detector drifts in terms of its calibration with respect to the other detector, the error in
the projection measurement Pj will be propagated throughout the backprojection
process, leading to a ring artifact (Fig. 10-73B). With the fourth-generation CT, a
detector fan geometry is used (Fig. 10-73C). The same detector is used to make both
measurements necessary to compute the projection value, and therefore any drift
in the detector will be intrinsically factored out since the Ir and Ij measurements are
made by the exact same detector element. This fundamentally eliminates the production of ring artifacts in the reconstructed CT image.
Fourth-generation CT geometry was devised in the 1970s as a method to combat
ring artifacts with third-generation CT. Since that time, however, there have been
enormous improvements in detector stability and in algorithms that monitor and
correct for electronic gain fluctuations in the detector hardware. In addition to these
preprocessing steps to reduce ring artifacts, algorithms have been developed that
can correct for ring artifacts in the reconstructed CT images. These correction procedures have all but eliminated the ring artifact problem in modern third-generation
CT systems.
374
Suggested Reading
Boone JM. Determination of the pre-sampled MTF in computed tomography. Med Phys 2001;28:
356360.
Defrise M, Townsend D, Geissbuhler A. Implementation of three-dimensional image reconstruction for
multi-ring positron tomographs. Phys Med Biol. 1990;35(10):13611372.
Feldkamp LA, Davis LC, Kress JW. Practical Cone-beam algorithm. JOSA A. 1984;1:612619.
Joseph PM, Spital RD. A method for correcting bone induced artifacts in computed tomography scanners.
J Comput Assist Tomogr 1978;2(1):100108.
Kalender WA. Computed Tomography: Fundamentals, System Technology, Image Quality, Applications, 3rd
Edition, Publics Publishing, Erlangen, Germany 2011.
Kalender WA, Seissler W, Klotz E, Vock P. Spiral volumetric CT with single-breath-hold technique,
continuous transport, and continuous scanner rotation. Radiology. 1990 Jul;176(1):181183.
Seidensticker PR, Hofmann LK. Dual source CT imaging. Heidelberg, Germany: Springer, 2008.
Chapter
11
X-ray Dosimetry in Projection

Imaging and Computed
Tomography
11.1 Attenuation of X-rays in Tissue
When x-rays are used to image a patient, the primary x-ray beam is incident upon
the entrant tissues of the patient and is attenuated at depth. For a depth x of
tissue, the attenuation of the primary x-ray beam is determined by

D ( x ) = Do e eff x
[11-1a]
where Do is the dose at the skin surface, D(x) is the dose at greater distances from the
skin, and meff is the effective linear attenuation coefficient. The effective linear attenuation coefficient is that value that most closely matches the attenuation profile of a
polyenergetic x-ray spectrum. Whereas Equations 11-1a and 11-1b is suitable for
most cases, a more complete mathematical description of primary x-ray beam attenuation for a narrow beam geometry when an x-ray spectrum f(E) is used is given by
E max
D( x ) =
dE (E ) e ( E ) x ,
[11-1b]
E = E min
Figure 11-1 illustrates the intensity of the x-ray beam as a function of tissue thickness
of the patient. X-ray beams are attenuated exponentially (i.e., Equations 11-1a and
11-1b), and this is plotted on a log-linear graph (Fig. 11-1). At 80 kV (2.5 mm added
Al), only 1 out of 118 photons survives to a depth of 20 cm, and only 1 out of 353
photons survives to 25 cm. At 120 kV, only one out of 190 x-ray photons in the beam
penetrates a 25-cm thickness of tissue and exits the patient toward the detector. These
numbers illustrate that primary x-ray photons that exit the patient are relatively rare,
and therefore structures that attenuate postpatient primary photons (patient table,
detector covers, anti-scatter grid, etc.) should be designed to be minimally attenuating.
As a primary x-ray beam strikes an object and undergoes attenuation, many of
the interaction events occur as Rayleigh or Compton scattering. Equations 11-1a and
11-1b describe the attenuation of primary x-rays but do not and cannot accurately
describe what happens to the scattered x-ray photons that occur from primary photon scattering interactions. It has been observed that scattered photons can undergo
subsequent scattering events. To accurately account for scattered photon dose deposition, Monte Carlo techniques, described later, are necessary.
Although attenuation leads to the deposition of radiation dose in the patient,
attenuation is necessary for creating a projection image. In the absence of attenuation, no dose would be deposited but there would be no useful image either. In
diagnostic radiology, x-rays interact by a combination of complete absorption events
(photoelectric absorption) and scattering events (Compton and Rayleigh scattering).
375
376

1.000000
60 kV
80 kV
0.100000
Relative X-ray Fluence
100 kV
120 kV
0.010000
0.001000
0.000100
0.000010
0.000001
0
10
15
20
25
30
35
Tissue Thickness (cm)
40
45
50
FIGURE 11-1 The attenuation profiles for x-ray spectra at four different tube voltages are
shown. The vertical axis is logarithmic, which compresses the actual magnitude of the attenuation.
Only the photoelectric effect and Compton scatter lead to dose deposited in tissue.
Scattered radiation has an important role in redistributing the primary beam dose
deposition patterns, and consequently, tissues outside the collimated (primary) x-ray
beam receive some radiation dose due to scattered radiation. To the extent that x-ray
scattering results in a fraction of x-ray photons leaving the patients body, scatter
represents a physical mechanism by which radiation dose to the patient is slightly
reduced. However, in imaging procedures such as fluoroscopy where staff are near
the patient, the scattered x-rays leaving the patient then become a source of radiation dose for the personnel in the fluoroscopic suite. All of the photon attenuation
mechanisms mentioned above, as well as many of the dose terms discussed below,
were introduced in Chapter 3.
Dosimetry in x-ray imaging is fundamentally related to x-ray attenuation, and the
effective penetration of the x-ray spectrum has an important role to play in regard to
attenuation (Fig. 11-1). The position of the x-ray field on the patient, the size of the
x-ray field, the thickness and composition of the patient along the direction of the
x-ray beam, and whether or not the x-ray beam changes position during the acquisition (as with tomography) are the patient and geometrical factors that influence
overall dose and the pattern of dose deposition in the patient.
Absorbed dose, or simply dose, is defined as

Absorbed Dose =
Energy Imparted
Mass
[11-2]
When Energy Imparted has the units of joule, and Mass is in kilograms, the dose unit is
the gray (joule/kg). Thus, a millijoule deposited in a gram of tissue leads to a gray of
dose. One gray is equal to 100 rads, and most (but not all) x-ray imaging procedures
result in absorbed doses from tenths of mGy to a few tens of mGy.
Over the years, there have been a number of dosimetry metrics that have been
used to estimate the radiation dose to patients for x-ray imaging. As a result, there
are several different surrogate dose metrics that the practitioner should be aware of,
in addition to absorbed dose. This chapter outlines some of the most commonly used
figure of merits, as well as currently accepted dosimetry methods.
Chapter 11 X-ray Dosimetry in Projection Imaging and Computed Tomography
377
11.2 Dose-Related Metrics in Radiography and Fluoroscopy

For a projection image such as an abdominal radiograph, the magnitude of the dose
changes dramatically with depth in the patient. This huge heterogeneity in dose
makes an accurate dose assessment difficult. Figure 11-2 underscores this. Even
across the distance of a fist-sized organ such as the kidney, the dose can change by as
much as a factor of 3. In order to estimate organ doses, the exponential nature of the
dose distribution must be taken into consideration.
While it is relatively difficult to compute various tissue doses in projection radiography, it is relatively easy to estimate the kerma at the skin surfacethe entrance
skin kerma (ESK). In order to estimate the entrance skin dose (ESD), the air kerma
at the skin layer must be estimated.
X-ray Tube Output: mGy per 100 mAs

Most periodic quality control procedures performed on radiographic systems involve
the measurement of the air kerma as a function of mAs at a specified distance from
the x-ray source, over a range of kVs. Typically, the air kerma (mGy per 100 mAs at
100cm) is measured (in the past, exposure in the units of mR per mAs was measured).
An example of such information is shown in Figure 11-3. The output values (also listed
in Table 11-1) are valid for the specific x-ray system in which the data were measured,
and other x-ray systems will have slight-to-substantial variations from these values
depending primarily on the amount of inherent filtration there is in the x-ray beam.
X-ray Imaging Geometry: Source-to-Skin Distance (SSD)

Referring to Figure 11-4, the distance from the x-ray tubes focal spot to the entrance
skin layerthe source-to-skin distance (SSD)needs to be estimated. Since most
2.00 mGy
0.66 mGy
0.22 mGy
0.07 mGy
0.02 mGy
FIGURE 11-2 This figure shows the typical attenuation values as the x-ray beam penetrates the tissues in
the body, represented here by an axial CT image. The x-ray dose at the surface of the patient is 2.0 mGy, and
this value is attenuated to just 0.02 mGy at the exit surface, demonstrating that the x-ray intensity is reduced
by a factor of 100 in this example.
FIGURE 11-3 The output kerma for

a typical x-ray tube is illustrated. X-ray
tube output efficiency increases slightly
with tube voltage, as the upward curve
of this graph illustrates.
Kerma @ 1 meter (mGy/100 mAs)
378
18
16
14
12
10
8
6
4
2
0
40 50 60 70 80 90 100 110 120 130 140 150 160
tube voltage (kv)
radiography is performed at a source-to-detector distance of 100 cm (chest radiography at 183 cm), the SSD can be estimated by knowing the approximate thickness of
the patient. For example, for a 25-cm-thick patient, with a 2-cm separation between
the patient and the detector, the SSD is computed as SSD = 100 [25 + 2] = 73 cm.
TABLE 11-1HALF-VALUE LAYER AND OUTPUT LEVELS
AS A FUNCTION OF kV FOR A TYPICAL

GENERAL DIAGNOSTIC X-RAY SYSTEM
kV
HVL (mm Al)
OUTPUT (mGy per 100 mAs)
40
1.15
1.1
45
1.36
1.7
50
1.57
2.3
55
1.79
2.9
60
2.00
3.5
65
2.22
4.3
70
2.43
5.0
75
2.65
5.8
80
2.86
6.7
85
3.08
7.5
90
3.29
8.5
95
3.50
9.5
100
3.71
10.5
105
3.92
11.6
110
4.12
12.7
115
4.32
13.9
120
4.51
15.1
125
4.71
16.4
130
4.89
17.8
135
5.08
19.2
140
5.25
20.6
Output was measured free in air at a distance of 100 cm from the x-ray source,
along the central beam. Multiply output column by 1.45 to get mR/mAS.
379
FIGURE 11-4 The standard imaging geometry in radiography is illustrated, here for an anterior-posterior (AP) projection image. The source to detector distance (SID) is typically
100 cm for radiography, except it is 183 cm for upright
chest radiography. The source to skin distance (SSD) can be
estimated by knowing the SID, the thickness of the patient,
and the dead space between the cover of the detector and
the actual detector surface (air gap in this figure).
SSD
SID
thickness
Air Gap
Estimation of Entrance Skin Kerma

Let us say that a radiograph was acquired using 80 kV and 82 mAs, and the SSD in
Figure 11-4 was estimated to be 73 cm. Table 11-1 shows that at 80 kV, the output
of this x-ray system is 6.7 mGy per 100 mAs at a distance of 100 cm from the source.
The x-ray tube output at a given kV is linearly proportional to the mAs. For example,
double the mAs and the air kerma will double. For the 82 mAs used in the example,
the output kerma K is

K 82mAs =
82
K 100mAs
100
[11-3a]
where K100mAs is determined from Table 11-1 for 80 kV (6.7 mGy per 100 mAs), and
thus, the kerma for this technique at 100 cm is 5.5 mGy. Correction for the inverse
square law is necessary to compute the entrance kerma at the 73-cm SSD:
2
100 cm
ESK =
K 82 mAs
73 cm
[11-3b]
As determined from Equation 11-3a, K82mAs at 100 cm was 5.5 mGy. The inverse
square correction in this example equates to a factor of 1.88, resulting in an ESK of
10.3 mGy (1.18 R).
Calculations as described above make use of so-called free-in-air measurements
of kerma, and thus, x-ray scattering in tissue does not contaminate the estimate.
Free-in-air measurements are also known as without backscatter. If one were to
actually measure the kerma by placing an ionization chamber in front of a tissueequivalent phantom, the probe would measure both the incident primary beam and
a considerable amount of scattered radiation from the phantom; because this scatter
is directed back toward the x-ray source, it is called backscatter. When backscatter is
included, depending on the beam energy and geometry, the measurement would be
380
about 15% to 30% higher than a free-in-air measurement at the same position and
with the same x-ray technique factors. It is common to refer to ESK measurements as
with backscatter or without backscatter, to clarify the measurement methodology.
In general, most tabulated dose data (but not all) are computed for free-in-air measurements, as will be discussed below.
The ESK (formerly the quantity entrance skin exposure was used, involving different units) is a surrogate measure of the dose used in projection radiography. The
ESK is not an accurate estimate of tissue dose but has evolved as an informal metric
for comparing doses among radiographic procedures. For example, in the example of abdominal radiography given above, the ESK was computed as 10.3 mGy.
A chest radiographic procedure may have an ESK of about 0.09 mGy or 90 mGy,
but this is one of the lowest dose radiographic procedures performed. The abdominal radiograph requires more kerma because the abdomen (mostly solid tissue) at
typically80kV attenuates considerably more than the lung parenchyma in the chest
radiograph at 120 kV.
Estimation of Skin Dose in Radiography

The ESK represents the radiation intensity as the x-ray beam impinges on the surface
of the skin. The skin dose, at the very surface of the skin where no attenuation has
occurred, is computed as
en

tissue
ESD =
ESK
en

air
[11-4]
Equation 11-4 is strictly correct only for a narrow entrance beam of primary photons. As the x-ray beam striking the surface of the skin gets larger, the role of scatter
increases and for large beams a backscatter factor should be included. The ratio of
the mass energy attenuation coefficients in Equation 11-4 has a numerical value of
about 1.06 for general diagnostic x-ray energies.
Skin Dose in Fluoroscopy

The computation of skin dose in fluoroscopy uses Equation 11-4, but slightly different methods are used to estimate ESK. In fluoroscopy, the kerma rate is a function of the mA and kV, and the mA fluctuates during the procedure depending on
the positioning of the fluoroscopic system relative to the patients body and on the
mode settings of the fluoroscopy system. The automatic brightness control system on
all modern fluoroscopy systems modulates the mA based on the radiation intensity
striking the fluoroscopic detector. A common way to estimate this is to use the setup
shown in Chapter 9 (Fig. 9-15). Various thicknesses of an attenuator such as polymethyl methacrylate (PMMA) are positioned in the beam as the kerma rate is measured
at the detector surface. The kerma rate is measured as a function of the thickness of
the PMMA over a range of possible patient thicknesses, and these measurements can
be used to estimate the entrance kerma rate to the patient, as shown in Figure 11-5.
If the chamber is positioned close to the scatter material, then the measurements will
be with backscatter. Alternatively, the radiation meter can be positioned well away
from the attenuator (i.e., closer to the x-ray tube), which would reduce the contribution of scatter to the measurement such that it is without backscatter. In this later
381
15 frames per second - fiat panel - lucite x 1.19 to water equivalence
Air Kerma Rate (uGy / sec)
900
800
700
48 cm FOV
600
27 cm FOV
15 cm FOV
500
400
300
200
100
0
10
15
20
25
30
35
Patient Thickness (cm)

FIGURE 11-5 In fluoroscopy, the measured x-ray flux at the detector drives the automatic brightness control
circuitry, and the air kerma rate is controlled by the mA and kV of the system. In-field measurements of patient
dose make use of plastic phantoms of varying thickness. The field of view also affects the gain of the detector
system, and so measurements are made as a function of field of view. The data shown in this plot are for a flat
panel detectorbased fluoroscopic system.
case, the inverse square law is then used to calculate the measured kerma rates at the
entrant surface of the patient.
Modern fluoroscopy systems have the ability to automatically record the total
fluoroscopy time used (in minutes) for a procedure, and the total fluoroscopy time
is converted to seconds and multiplied by the kerma rate to determine the total ESK.
From this, Equation 11-4 can then be applied to compute ESD. This calculation
assumes that the entrance beam is located at the same position on the patient, such
that the same patch of skin was exposed during the entire fluoroscopic procedure.
When the fluoroscopy system is positioned at a number of different positions such
that different areas of skin receive the entrance beam (as when both PA and lateral
projections are used), the total time for each position must be estimated for the calculation of ESD in each location.
Many fluoroscopic systems employ Kerma-Area-Product (KAP) meters. These
devices are mounted between the x-ray source and the patient (Fig. 9-15), and
monitor the total KAP during a fluoroscopic procedure. If radiographic images are
acquired, the doses from these are also included in the KAP value. The value of KAP
is the product of the total air kerma exiting the x-ray tube housing and the area of
the beam, and hence the units are mGy-cm2. A calibration procedure can be performed to convert the KAP value into an estimate of ESK. Newer KAP meters also are
required to provide a center point kerma reading. This value can be used directly to
estimate the ESK, with the caveats mentioned above in regard to variable positioning
of the beam on the patient. Because of the relative consistency of output of modern
fluoroscopic systems combined with the comprehensive digital design of the systems
(i.e., where SSD is known), for a well-engineered system, the ESK can be inferred
(estimated), forgoing the need for a physical KAP meter.
The estimate of skin dose is useful primarily when fluoroscopic, cinefluorographic, and fluorographic procedures are performed and combine to produce high
skin doses with potential deterministic effects. Deterministic effects in fluoroscopy
are rare and usually occur in only the most lengthy procedures involving complicated
382
cardiac or vascular interventional procedures. Estimates of skin dose will help to

gauge whether dermatological effects such as epilation or erythema may occur, and
to determine if the patient should be monitored for such effects. While the skin dose
is the highest dose in the patient for projection radiographic procedures such as
fluoroscopy and radiography, it is not an accurate measurement of the entire patient
dosethat is discussed in the next sections.
11.3 Monte Carlo Dose Computation

Most modern x-ray dosimetry is based upon Monte Carlo calculations, with reference
to the stochastic nature of gambling in the principality of Monaco. Monte Carlo.
Monte Carlo procedures use computer simulation to study the dose deposition during an x-ray exposure. Computer programs can be used to produce random numbers, similar conceptually to flipping a coin, except that the output of the random
number generator is usually a continuous variable between 0 and 1. Using these values, any probability distribution can be produced. Using the known physics of x-ray
interaction in tissues, the known probabilities of scattering events (both Rayleigh and
Compton scattering) as a function of angle and x-ray energy, a geometrical model of
the simulated patient, and the random number generator, Monte Carlo calculations
compute the three-dimensional dose distribution for a given exposure geometry
(Fig. 11-6). The power of modern computers allows Monte Carlo simulations to be
performed using many simulated photons, which reduces statistical uncertainty in
the computation. Typically, 106 to 108 photons are simulated and their individual trajectories and interactions are tracked by the Monte Carlo program, photon by photon and interaction by interaction. Even though millions of photons (i.e., 106) can be
Backscattered
photon
Scattered photon
Primary photon
Dose distribution
FIGURE 11-6 Monte Carlo dosimetry is based on tracking the trajectory and energy deposition of x-ray
photons through a mathematical phantom with a defined geometry and composition. Millions to billions of
photons are typically simulated to produce a computed dose distribution in the phantom; using realistic mathematical models of patient anatomy, organ doses can be estimated with excellent accuracy.
383
TABLE 11-2ORGAN CONVERSION FACTORS FOR SELECTED RADIOGRAPHIC
PROCEDURES
Gy PER mGy ENTRANCE kerma
HVL FOR ENTRANT X-RAY BEAM (mm Al)
1.0
2.0
3.0
4.0
5.0
Lungs
2.4
9.6
17.2
24.0
28.6
Active bone
marrow
7.4
30.9
59.5
85.9
109.9
Thyroid
0.0
0.1
0.2
0.3
0.3
Trunk tissue
68.7
137.4
187.8
224.4
253.0
Ovaries
59.5
183.2
296.6
385.9
453.4
Uterus
83.6
249.6
395.0
506.1
586.2
Adapted from Table 41 (AP Abdominal Radiography) 100 cm SID, 35 43 cm field of view, in Handbook of
Selected Tissue Doses for Projections Common in Diagnostic Radiology, HEW publication (FDA) 89-8031 (1989).
simulated in Monte Carlo programs, the number of simulated photons is still much
lower than that which occurs in a physical radiographic acquisition. For example, for
a typical x-ray beam used in chest radiography (a very low dose examination), a total
of about 1.6 1011 photons are incident upon the patient. For Monte Carlo calculations, the organ doses are normalized per incident kerma so that the results can be
calibrated to the photon intensities used in actual imaging procedures.
Typically, Monte Carlo calculations are performed over a range of x-ray energies (e.g., 10 to 140 keV by 1 keV), and then the results are normalized for a given
x-ray spectrum. X-ray spectra can be generated by computer models, or tabulated
measured spectra can be used. Energy deposited in tissue from x-ray interactions is
tallied when a photoelectric interaction occurs (in such cases, essentially all of the
x-ray energy is deposited locally in tissue), or when a Compton scattering event takes
place (the energy deposited at the site of a Compton interaction is the difference in
energy between the incident x-ray photon and the scattered photon). Rayleigh scattering results in a slight redirection of the x-ray photon, but no energy is absorbed
and so no dose is conveyed. After millions or even billions of x-ray interactions are
simulated, the deposited energy is combined with the known mass of each small tissue volume and the absorbed dose can then be computed.
For radiography, Monte Carlo simulations have been performed for a comprehensive set of standard imaging projections, such as the chest radiograph (AP and
lateral), skull (AP and lateral), and so on. Typically, the doses to the individual organs
in the mathematical model are computed individually. A number of publications are
available that contain tables that describe the organ dose for each radiographic projection. The data tables are provided over a range of x-ray spectra or beam qualities
for each projection. An example for abdominal radiography is given in Table 11-2.
11.4 Equivalent Dose

In industries such as nuclear power or isotope projection facilities, particles such
as neutrons, alpha particles, and energetic ions can be sources of radiation dose.
Various types of particulate radiations at different energies do not produce the
384
same biological damage per unit absorbed dose as x-ray or gamma-ray photons.
The relative biological effectiveness of these heavy charged particle radiations to that
of x-ray or gamma photons (and the energetic electrons they produce) is represented
by the radiation weighting factor wR. When the absorbed dose D to a tissue is known,
the equivalent dose H can be computed using

H D wR
[11-5]
When the units of D are milligray (mGy), and the unit of H are millisievert (Sv), the units
of wR are mSv mGy1. Biological effects are discussed in more detail in Chapter 20.
The radiation weighting factor wR for x-rays and g-rays is 1.0, and although the
equivalent dose can in principle be computed with x-rays or g-rays using this factor of
1.0 (resulting in the units of mSv), this practice is discouraged in x-ray dosimetry as
it can lead to confusion with effective dose (E), which is discussed below.
11.5 Organ Doses from X-ray Procedures

Because the radiosensitivity of each organ is different, the computation of organ doses
for x-ray dosimetry is considered the most accurate way in which to assess the overall
radiation dose. A comprehensive understanding of organ doses also allows the radiation
risks to be estimated for the generic patient. Dose assessments to the generic patient do
not take into consideration patient age, gender, or body size. A complete assessment
of organ doses also allows one to compute cancer detriment or E, as discussed later. A
list of organs and tissues that affect the radiation risk profile is provided in Table 11-3.
TABLE 11-3TISSUE WEIGHTING FACTORS (wt)
FOR VARIOUS TISSUES AND ORGANS

(ICRP 103)
TISSUE
Wt
Gonads
0.08
Bone marrow
0.12
Colon
0.12
Lung
0.12
Stomach
0.12
Bladder
0.04
Breast
0.12
Liver
0.04
Esophagus
0.04
Thyroid
0.04
Skin
0.01
Bone surface
0.01
Brain
0.01
Salivary glands
0.01
Remainder
0.12
Total
1.00
Wt values are in the units of mSv/mGy.
385
This list of organs clearly is biased toward the visceral organs in the abdomen, and
consequently, for the same absorbed dose levels, abdominal x-ray imaging is associated with higher risks than for similar radiation dose levels to the head or lower
extremities.
Organ doses can be estimated in radiography or fluoroscopy using a series of
Monte Carloderived tables for each radiographic projection, and similar concepts
hold for computed tomography as well. These will be discussed below.
11.6 Effective Dose

A comprehensive estimation of organ dose is the first step in a complete description
of stochastic risk to a generic patient. Once organ doses to each of the organs listed
in Table 11-3 are computed (in many imaging procedures, some organs will have
zero dose), then it is straightforward to compute the E. The equivalent dose concept discussed above described weighting factors wR, for different types of radiation,
but E uses tissue weighting factors (wT). Notice that in Table 11-3, the sum of the tissue weighting factors is 1.000, and the E concept can be thought of as a method of
computing radiation detriment for partial body exposures, typical of medical imaging proceduresfor example, if a patient received X amount of radiation dose to a
tissue T, the whole-body dose that has the same risk would be calculated as wT X.
Conversely, if a person were to receive a homogeneous whole body x-ray radiation
exposure where all organs received the same absorbed dose D, then the E would be
E = 1.00 Sv/Gy D. The SI unit of E is the sievert (Sv). In medical imaging, the units
of effective dose typically encountered are mSv or mSv, and the units of obsorbed
dose D are mGy or mGy, thus wT has units of mSv mGy1.
The tissue weighting factors listed in Table 11-3 were defined by a scientific
committee (International Commission on Radiation ProtectionICRP 103), in their
evaluation of epidemiological data from radiation exposure in humans, primarily the
A-bomb survivors in Japan during World War II.
Cautionary Notes in the use of Effective Dose

The utility and interpretation of E can be complicated, and there are some necessary caveats that need to be understood in regard to E. First, despite common
practice (especially in CT), the use of E was never intended to be used for assigning
risk to a specific patientthat is, a specific patient who has undergone a particular
radiological procedure. The organ-specific radiation risks that are embedded in the
wT coefficients were calculated for a generic person, not a specific individual. Any
specific patient may have a completely different risk profile than a generic individualfor instance, the patient may be 5 years old, or he or she may be 85 years
old. The patient may be female, where risks are generally greater than for male
patients. The patient may weigh 40 or 140 kgeach with profoundly different
radiation doses and corresponding risk profiles than the average radiation worker.
The patient may also have had a disease such as cancer, where risks of cancer
induction are likely heightened by potential genetic predisposition chemotherapy,
radiation therapy.
Where E is useful in radiological imaging is for comparing various radiological imaging procedures. For example, the radiation risk between a chest radiograph
and a thoracic CT can be assessed, and the relative risk can be computed. Pediatric
head radiography can be compared to a modified low-dose head CT procedure for
386
e valuation of chronic sinus inflammationthere are many examples. Such a calculation may be useful in optimizing clinical imaging for a generic patient.
11.7 Absorbed Dose in Radiography and Fluoroscopy

The most straightforward way to compute the absorbed dose to a patient from a
radiographic procedure is to determine the ESK and x-ray beam quality, and then
locate the appropriate table for the radiographic projection (e.g., Table 11-2). These
tables were produced using Monte Carlo simulations. A series of simple calculations
then allows the computation of individual organ doses. Some tables provide direct
conversion from entrance surface dose to E for radiographic projections, and an
example of that for chest radiography is given in Table 11-4.
Since the introduction of the E concept in 1977, the organs selected for use in
computing E and the wT value assigned to them have changed twice (1991 and 2007)
as the ICRP has refined their models and updated their recommendations. Consequently, the usefulness of some of the E estimation tables may be limited by the E
formalism that was in effect at the time the table was created.
Other forms of direct dose estimation are possible. If one is interested in the point
dose from primary radiation at a certain depth d in the patient from a radiographic
projection, such as with fetal dose calculations, it is necessary to keep in mind that
the x-ray beam experiences both tissue attenuation and a reduction in intensity from
the inverse square law at depth. For an x-ray beam with an effective tissue linear
attenuation coefficient of meff, a known SSD, a known ESD (per Equation 11-4), and
a depth d beyond the skin surface (along the direction of the x-ray beam), the point
dose can be estimated as
Table 11-4
FILTRATION
(mm OF Al)
ANTEROPOSTERIOR
(mSv/mGy)
POSTEROANTERIOR
(mSv/mGy)
LATERAL
(mSv/mGy)
90
0.176
0.116
0.074
90
0.196
0.131
0.084
90
0.210
0.143
0.091
100
0.190
0.128
0.081
100
0.208
0.143
0.091
100
0.222
0.155
0.098
110
0.201
0.139
0.088
110
0.219
0.154
0.097
110
0.232
0.165
0.104
120
0.211
0.149
0.094
120
0.228
0.163
0.103
120
0.240
0.174
0.110
X-RAY POTENTIAL (kV)
Note: This table is an example for chest radiography, where the ED (msv) per entrance surface dose (mGy) is
given for three different radiographic projections. The kV and filtration levels allow adjustment for different
x-ray technique factors used for the procedure. Effective dose per entrance surface dose (mSv per mGy) for
chest radiology 183 cm SID.
Source: From Estimation of effective dose in diagnostic radiology from entrance surface dose and dose-area
product measurements. National Radiation Protection Board of Great Britain, 1994, Table 7.
387
SSD - eff d
Dd =
e
ESD
SSD + d
[11-6]
When the dose from scattered radiation is not considered, as in the above example,
the estimates should be considered accurate to within an order of magnitude. The
most common method to accommodate for scattered radiation is to use Monte Carlo
simulation techniques or tables derived from them.
11.8 CT Dosimetry and Organ Doses

CT dosimetry is in a time of flux. As CT scanners have become more complicated
and more capable, there are numerous operational modes that place challenges on
the accuracy of dosimetry. Herein, a basic overview of CT dosimetry is provided.
Organ Doses in CT
Organ dosimetry in computed tomography is commonly performed using table
lookup as with radiography; however, the normalization process is not the same. In
radiography and fluoroscopy, the ESK is computed and used with the tables to compute organ dose. In CT, the concept of ESK is ambiguous due to the rotation of the
x-ray source around the patient. The solution in this case is to compute the air kerma
at the isocenter of the scannerthat is, at the center of rotation of the gantry with
nothing in the beam. Most tables for CT organ doses are normalized to the air kerma
at the isocenter. As with radiography, these tables are typically generated using Monte
Carlo methods. There are a number of commercially available dose computation
packages that are used by medical physicists to expedite organ dose computation in
CT. It is very important to remember that organ dose computation in CT using tables
generally assumes that the entire organ has been irradiated. For CT angiography and
perfusion studies and other CT acquisition protocols, this may not be the case. Dose
in CT angiography and perfusion is discussed below.
Due to the rotational irradiation geometry used in CT, the radiation dose distribution in the patient is far more homogeneous than in radiography or fluoroscopy.
The use of a beam-shaping filter further reduces heterogeneity in the dose distribution. Thus, in CT, the dose gradients are very slight, and the distribution depends on
the diameter and shape of the patient and on the beam quality (kV).
In helical (spiral) acquisition, most modern CT scanners have dose modulation
modes that adjust the x-ray tube output (by varying the mA) as the gantry rotates
around the patient and as the table translates the patient through the rotating x-ray
beam. Dose modulation modes generally increase the mA (and hence dose rate) as
the x-ray beam is aligned along thicker x-ray paths through the patient, and the
mA is reduced for thinner x-ray path lengths through the patient. The changing
mA in dose modulation mode slightly reduces the accuracy of table lookup CT
dosimetry.
Computed Tomography Dose Index, CTDI

The computed tomography dose index (CTDI) was originally designed as an index,
not as a direct dosimetry method for patient dose assessment. Over the years, there
have been enhancements and modifications to the original CTDI concept that have
attempted to make it a more accurate patient dosimetry method, with mixed results.
388
10 mm pencil chamber
PMMA plug
peripheral hole
32 cm body PMMA phantom
center hole
16 cm head PMMA phantom

FIGURE 11-7 The computed tomography dose index (CTDI) is measured using either a 16-cm or 32-cmdiameter polymethyl methacrylate (PMMA) phantom. The dosimeter is placed serially in the center hole and the
peripheral hole, and the measurements are combined to produce the weighted CTDI, as described in the text.
Despite limitations that will be discussed later, CTDI-based dosimetry is the current
worldwide standard for estimation of patient dose in CT.
The basic CTDI100 measurement involves the use of a 100-mm-long cylindrical (pencil) chamber, approximately 9 mm in diameter, inserted into either the
center or a peripheral hole of a PMMA phantom. There are two standard PMMA
dosimetry phantoms; the body phantom (Fig. 11-7) is 32 cm in diameter and 15
cm long, and the head phantom is 16 cm in diameter and 15 cm long (Fig. 11-7).
The head phantom also serves as a pediatric torso phantom. With the pencil chamber located at the center (in the z-dimension) of the phantom and also at the center
of the CT gantry, a single axial (also called sequential) CT scan is made (using no
table translation).
An ionization chamber can only produce an accurate dose estimate if its
entire sensitive volume is irradiated by the x-ray beam. Therefore, for the partially irradiated 100-mm CT pencil chamber, the nominal beam width (i.e., the
total collimated x-ray beam width as indicated on the CT console) is used to correct the chamber reading for the partial volume exposure. The 100-mm chamber
length is useful for x-ray beams from thin slices (e.g., 5 mm) to thicker beam
collimations such as 40 mm. The correction for partial volume is essential and is
calculated using

K corrected
100 mm
K measured
B
[11-7]
where B is the total collimated beam width, in mm, for a single axial scan. The value
of B is the product of the width of an individual CT detector projected to the scanners isocenter (T) and the number of active detectors, n, and so typically B = nT. For
example on a 64 channel CT scanner (n = 64) with each detector channel measuring
0.625 mm (T = 0.625 mm), B = 64 0.625 mm 40.0 mm.
389
The CTDI100 is defined as

CTDI 100 =
1
nT
+50 mm
D ( z ) dz
[11-8]
L = 50 mm
The CTDI100 in the above equation describes the measurement of the dose distribution, D(z), along the z-axis, from a single circular (axial or sequential) rotation of the
scanner with a nominal x-ray beam width of nT, where the primary and scattered radiation are measured over a 100-mm length, and where the center of the x-ray beam is
at z = 0. The nominal beam width refers to the beam width as reported by the scanner,
not the actual measured beam width, which is generally slightly wider. As mentioned
previously, CTDI100 measurements are made for both the center (CTDI100,center) and
periphery (CTDI100,periphery). Combining the center and peripheral measurements using
a 1/3 and 2/3 weighting scheme provides a good estimate of the average dose to the
phantom (at the central CT slice along z), giving rise to the weighted CTDI, CTDIw:

CTDI w 1 3 CTDI 100,center 2 3 CTDI 100, periphery
[11-9]
In helical (also called spiral) CT scanning, the CT dose is inversely related to the
helical pitch used, that is,
dose
1

pitch
[11-10]
where the pitch is defined as the table translation distance (mm) during a full rotation
(360 degrees) of the gantry, divided by the nominal beam width nT (in mm). Because
of this relationship, the CTDIw is converted to the volume CTDI (CTDIvol) using

CTDI vol =
CTDI w

pitch
[11-11]
Most scanners have the ability to display the CTDIvol on the CT scanner console
prior to the actual scan. The value can be displayed because the CT manufacturer has
measured CTDIvol in the factory over the range of kV values for that model of scanner,
and then that stored value, scaled appropriately by the mAs and pitch, is displayed
on the console.
The product of the CTDIvol and the length of the CT scan along the z-axis of the
patient, L, is the dose length product (DLP):

DLP = CTDI vol L
[11-12]
It has been shown that the DLP is approximately proportional to effective dose
(E). The slope of the E versus DLP relationship is called the k value, and there are
different k values for various CT examinations, as given in Table 11-5.
Limitations of CTDIvol
The most important limitation of CTDIvol is that it is a dose index, and was not initially meant to be a measurement of dose per se. CTDI concepts were meant to enable
medical physicists to compare the output between different CT scanners and were not
390
TABLE 11-5CONVERSION FACTORS (k FACTORS)
FOR ESTIMATION OF EFFECTIVE DOSE (IN

mSv) FROM DOSE-LENGTH PRODUCT (IN
mGy-cm), FOR VARIOUS CT EXAMINATION
TYPES (from AApm report 96)
CT EXAM TYPE
k FACTOR (mSv/mGy-cm)
Head
0.0021
Chest
0.014
Abdomen
0.015
Abdomen-pelvis
0.015
Pelvis
0.015
o riginally intended to provide patient-specific dosimetry information. The body CTDIvol

as reported by the CT scanner, or as measured on a CT scanner, is a dose index that
results from air kerma measurements at two locations to a very large (32-cm diameter)
cylinder of PMMA plastic with a density of 1.19 g/cm3. In terms of human dimensions,
the 32-cm-diameter PMMA body phantom corresponds to a person with a 119 cm (47)
waistlinea large individual, indeed. For smaller patients, the actual doses are larger
than the CTDIvol for the same technique factorsthus, the CTDIvol tends to underestimate dose to most patients. In light of this limitation, researchers have recently shown
that patient size conversion factors can be used with the CTDIvol reported on a given
CT scanner to produce size specific dose estimates (SSDEs). These conversion factors
for the CTDIvol measured on a 32-cm-diameter phantom are illustrated graphically in
Figure 11.8. The conversion factors described in AAPM Report 204 may be used to
more accurately estimate size-corrected doses from the CTDIvol, and these conversion
factors are (to a first approximation) independent of scanner manufacturer and tube
voltage. However, some CT scanner models use the CTDIvol value measured using the
16-cm-diameter PMMA phantom, and caution is needed to ensure that the correction
factors specific to the appropriate reference phantom are used.
Another fundamental limitation of the CTDIvol is that it is computed, as defined in
Equations 11-8 to 11-11, as the dose (in air) at the center of a 100-mm-long CT scan.
However, because scattered radiation is a considerable component of radiation dose to
3.5
Normalized Dose Coefficient
FIGURE 11-8 This figure shows data

taken from AAPM Report 204. This curve
shows the conversion factors as a function
of the effective diameter of the patient,
relative to the 32-cm-diameter CTDIvol phantom. The data from this curve can be used
to adjust the displayed CTDIvol when the
effective diameter of the patient is known.
More detailed information is provided in
AAPM Report 204.
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0
10
15
20
25
30
35
Effective Diameter (cm)
40
45
50
391
the patient in CT, the scattered dose distribution is important. The dose from scattered
radiation in CT tends to be most intense along the z-axis close to the primary beam,
and decreases as the distance along z from the primary CT beam increases. The notion
of the dose spread function, similar to the line spread function, has been described,
and dose spread functions are illustrated in Figure 11-9. The important observation
in this figure is the fact that the scatter tails are low-amplitude, long-range functions
that extend significant distances in the z-dimensions. Because of this, the dose tends
to build up (both inside and outside the primary CT scan volume) as the scan length
of the CT scan increases. The dose at the center (along z) of the scanned region in the
patient builds up the most, and so the dose at the edge of the scanned field is smaller
than the dose in the center. These trends are shown in Figure 11-10, where the dose
profiles in the z-dimension are illustrated for a number of different CT scan lengths.
The amplitude of the primary beam was the same for all dose profiles shown in Figure
11-10, and the dramatic increase in dose at the center of the field of view illustrates
the consequence of summing those low-amplitude, long-ranged scatter tails.
As the scan length approaches the range of the scattered radiation tails, the dose
at the center of the scan no longer increases and the dose at the center then becomes
asymptoticit no longer increases as the scan length increases. This is shown in
Figure 11-11, which is a graph of the dose at the center of the scanned field of view as
a function of scan length (or beam width in cone beam CT systems). This figure shows
1.0E+00
Relative Dose (AU)
50 cm
1.0E01
10 cm
1.0E02
30 cm
1.0E03
1.0E04
200
150
100
50
50
100
150
200
Position (mm)
Relative Dose (AU)
1.0E+00
140 kVp
1.0E01
80 kVp
1.0E02
200
150
100
50
50
100
150
200
Position (mm)
FIGURE 11-9 The long-range, low-amplitude tails from scattered radiation in the CT are shown in this figure.
A. the scatter tails are shown as a function of distance from the center (narrow) CT slice, for three different cylinder diameters. B. This curve shows that higher tube voltages generate scatter tails with greater range. At 200 mm
from the narrow beam primary source, the scatter intensity is about 1/100 compared to the center of the beam.
392

1.1
1.0
Relative Dose (a.u.)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
20
15
10
10
15
20
Position (cm)
FIGURE 11-10 The relative dose as a function of the position along the z-axis of a CT scan is shown, for
different scan lengths. Dose profiles for scan lengths of 5, 10, 20, and 30 cm are shown, and the dose at the
center of the field (see vertical dashed line) increases dramatically as the scan length increases.
that lower doses are delivered when the scan length is shorter than the equilibrium scan
length. Indeed, Figure 11-11 suggests that the known scan length of a specific CT procedure can be used to correct for the radiation dose value at the center of the beam.
The measurement procedure that is used to acquire the data shown in Figure11-11
is modified from the standard CTDI100 methods, and is described by the report of
AAPM Task Group 111 (TG-111). Instead of using a 100-mm pencil chamber and
measuring the dose to a partially irradiated chamber, a smaller chamber is used and
the helical (spiral) mode of the scanner is used to scan a given scan length, L, of the
scanner. A series of such measurements on scans of different scan lengths is performed
using the geometry illustrated in Figure 11-12. Using an integrating ion chamber
located at the center of each scan, each point on the curve can be measured using a
helical CT scan through a very long phantom. If a real-time radiation meter is used, the
measurement can be made with one helical scanthe real-time chamber is positioned
at the center of the field of view, and a helical scan is performed that interrogates the
1.0
0.9
Correction Factor
FIGURE 11-11 This figure shows the rise

to dose equilibrium curve. This curve is
essentially a plot of the data points along the
vertical dashed line shown in Figure 11-10,
as a function of scan length. The dose at
the center of the field increases and reaches
an asymptotic value once the scan length
(or beam width for cone beam CT systems)
approaches the width of the scatter tails.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
50 100 150 200 250 300 350 400 450 500 550
Beam Width or Scan Length (mm)
1.0
393
SL
Correction Factor
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
SL
0.1
0.0
0
50 100 150 200 250 300 350 400 450 500 550
Beam Width or Scan Lenght (mm)
SL
FIGURE 11-12 The rise to equilibrium curve can be measured using an integrating dosimeter, and CT scans
of different scan length are made to measure points along the curve. This figure illustrates AAPM Report 111
methodology.
entire length of the phantom (Fig. 11-13A). The probe measures a profile as shown in
Figure 11-13B, as a function of time. The velocity of the table is used to convert the
time axis of the probe read out to position along the phantom. The dose profile (Fig.
11-13B) shows a physical measurement of the scatter tails for a given set of imaging
parameters, and this curve can be integrated for various scan lengths to derive the rise
FIGURE 11-13 If a dosimeter capable of real-time read-out is used, only one acquisition is needed as the
phantom and x-ray chamber traverse (A) through the x-ray beam. The measured trace (B) shows the dose rate
as a function of position (idealized), and (C) shows the rise to equilibrium curve that can be determined from
the data in (B). This approach is the ICRU addendum to AAPM Report 111.
394
to dose equilibrium curve (Fig. 11-13C). The benefit of the ICRU approach shown
is Fig 11-13 is that only one scan through the phantom is required to determine the
entire equilibrium dose curve. The rise to dose equilibrium curve shown in Figure
11-11 essentially provides a series of correction factors that can be used for specific
scan lengths. Note that the correction factors would be dependent on scan parameters
and the data shown in Figure 11-11 should not be used for patient dosimetry.
Dose in CT perfusion
CT perfusion studies are used for evaluating vascular function in organs, primarily
the brain. Using a CT scanner with a reasonably wide coverage along the z-axis (such
as 30 to 40 mm), the scanner repeatedly interrogates the organ during the injection
of contrast agentscanning typically at 1 scan per second for 40 to 50 seconds.
The real-time imaging capability of the scanner is used with mathematical modeling
equations applied in software to derive a number of functional parameters in regard
to the imaged volume, including vascular permeability, time to peak enhancement,
blood flow rate, and blood volume. The application of CT perfusion studies is primarily for the assessment of stroke or vasospasm, and this procedure generally uses
no table movement and so the same tissue receives the radiation dose corresponding
to about 40 or more CT scans, all in one procedure. Some scanners have a shuttle
mode (see CT chapter for details), which allows the table to move back and forth
repeatedly, and this allows the assessment of tissue perfusion over a larger region of
tissue. The dosimetry methods pertinent to CT perfusion studies are different from
that of routine head CT, primarily because the entire brain is not scanned. When
the entire organ is not scanned (as with the brain in this procedure but this applies
in general), the tissue weighting coefficients, wt, are not accurate, and thus standard
effective dose calculations can be misleading. Concerns in CT perfusion studies are
primarily with the dose to the skin and the dose to the lenses of the eyes, and thus
other dosimetry methods will in general be required. The details of such methods are
beyond the scope of this text (see Bauhs 2008, Zhang 2011).
11.9 Computation of Radiation Risk to the Generic Patient

The estimation of radiation risk from a medical imaging procedure is typically computed by first computing the organ doses from the procedure. Organ dose assessment
is usually performed using a generic-sized patient, and the dose deposition is computed using Monte Carlo procedures based on the medical imaging procedure being
performed. Typically a mathematical phantom such as the MIRD* phantom is used,
and this phantom simulates each organ as a mathematical shape in a standard-size
body (Fig. 11-14). So-called voxelized phantoms have also been used, and these are
full CT scans of humans, where each organ has been outlined so that organ dose can
be estimated after the Monte Carlo procedure. In these Monte Carlo exercises, the
dose for each organ is computed, for various entrance x-ray beam geometries. Organ
doses are computed in the units of absorbed dose (mGy), usually normalized to ESK
in radiography or to air kerma at isocenter in computed tomography. Such tables
allow the estimation of organ doses for given radiological imaging procedures that use
different technique factors. For example, in Table 11-2 for abdominal radiography, the
organ dose coefficients for different beam qualities (half-value layers) are provided.
Neglecting the special risk issues from deterministic effects from very high dose
procedures such as fluoroscopically-guided interventional procedures, the risk from
395
FIGURE 11-14 The MIRD phantom is shown.

This phantom uses mathematically defined
organs and lends itself to Monte Carlo dosimetry simulations.
the vast majority of medical imaging examinations is the stochastic risk associated
with cancer induction. In the generic patient, which is assumed in the notion of
E, there is a risk coefficient that is used to weight each tissue type (Table 11-3), as
discussed previously. The product of the organ dose (in mGy) and the organ (or
tissue)-specific weighting factor is the E (in mSv) for that organ. The E for a given
medical imaging procedure is then the sum of all of the organ-specific E. The E from
a number of imaging procedures (such as may occur during a trauma workup) can
be summed across all procedures as well.
Instead of performing an organ-by-organ assessment of dose, as discussed in
the paragraph above, tables are provided in many reference publications that have
already performed this computation and these charts provide estimates of E for various radiological procedures (e.g., chest radiography, pelvic CT, and mammography).
For example, Table 11-4 provides coefficients for chest radiography, and this table
includes coefficients for various chest radiographic views (AP, PA, lateral) and for
different x-ray beam energies (half-value layers). Each value is normalized as the E
(mSv) per unit ESK (mGy). This allows the user to compute the E for the specific
technique factors used on the equipment that was used to image the patient. Table
11-5 illustrates the conversion factors for calculating E in CThere, the table lists
the E (mSv) as a function of the dose-length product (DLP, in mGy-cm) of the CT
scan. This difference in the normalization reflects the differences in the geometry
between CT and projection radiography.
Let us use the example of a thoracic CT scan, and compute the generic patient
dose using the E coefficient listed in Table 11-5. The hypothetical thoracic CT scan
was performed at 120 kV, 75 mAs, with a pitch of one and the scan length was 40 cm.
The reported CTDIvol on the scanner was 5.5 mGy, and the DLP was 219 mGy-cm.
Table 11-5 shows that for a chest CT, the k factor is 0.017 mSv per mGy-cm. Thus, the
ED for this study is computed as 219 (mGy-cm) 0.017 (mSv/mGy-cm) = 3.7 mSv.
396
Note that the conversion factor (0.017 mSv mGy1 cm1) did not take into account
the patients age, gender, or body size. Hence, this estimation is considered to be for
the generic member of the population.
The generally accepted risk (from BEIR VII) of a radiation-induced cancer fatality is 0.057 per 1000 mSv for a general population. So, the 3.7 mSv E from the CT
scan estimated in the above paragraph results in a risk of induced cancer death of
0.0002109 5 (0.057 [3.7/1000]), or one fatal cancer induced for every 4742
patients having this procedure.
11.10 Computation of Patient-Specific Radiation
Risk Estimates
Many scientists in the radiological community question the value of dose calculations for a specific patient, because every individual is different and has a unique
propensity for cancer induction, which likely varies from organ to organ and from
age to age. Calculations of cancer induction risk based on large cohorts are unlikely
to capture a specific individuals risks with accuracy. Nevertheless, as awareness of
the harms from ionizing radiation received during medical imaging procedures has
increased due to press coverage, patients or their physicians sometimes require an
estimate of dose from an imaging procedure. Beyond the radiation dose, the real
issue relates to what the risk may be from that imaging procedure.
As mentioned in the above section, the notion of E cannot specifically address
individual dosimetry. Using the example of CT, where this issue comes up most
often, the first step in risk estimation for an individual is to refine his or her physical
dose estimation by correcting the dose for the actual patient diameter (Fig. 11-8 and
AAPM Report 204) and also by adjusting the estimate for the actual scan length (Fig.
11-11 and related discussion).
Once the organ dose estimates are adjusted to the patient size and scan length,
then the Table 12D-2 in BEIR VII can be used to make relative risk calculations.
The age-dependent and gender-dependent incidence of radiation-induced cancer
deaths are illustrated in Figure 11-15. Here, the same case as mentioned in the
section above (5.5 mGy dose due to thoracic CT scan) will be used to demonstrate
age and gender dependencies in risk estimates. Computing the doses to each
organ, and then summing the cancer fatality rate for each organ, the relative risk
of this CT scan was estimated for male patients and was 0.00025 for a 10-yearold, 0.00012 for a 40-year-old, and 0.000063 for a 70-year-old person. Not surprisingly, the risk for a 10-year-old child is four times that of a 70-year-old, since
younger patients are likely to live longer than older patients and manifest a radiogenic cancer. For female patients, the risks were calculated as 0.000423, 0.00019,
and 0.0001 for the 10-, 40-, and 70-year-old patients, respectively. These differences in relative risk are reflected in Figure 11-15. The sensitivity of women
to radiogenic stochastic effects is greater than that of men, and the increases in
relative risk are estimated to be 1.62, 1.50, and 1.60 greater than male patients at
ages 10, 40, and 70.
Absolute risks are small numbers with many zeros ofter the decimal point.
Consequently, it is typical to invert these numbersfor example, a risk of 0.0010
equates to a risk of 1.0 per 1000. Table 11-6 provides these data for the example
given above. Notice that when age and gender are considered, the values for this
example range from risks of 1 in 400 to 1 in 2685. The concept of effective dose was
not used for these computations, which used the tables from BEIR VII (Table 11-7).
397
FIGURE 11-15 The cancer death rate (per 1000 individuals exposied to 10 mGy) is shown as a function of
age at irradiation, for men and women. Women and younger patients demonstrate higher cancer mortality
from the risks of radiation exposure. (Data taken from BEIR VII Table 12D-2.)
For the case of the generic patient discussed in the previous section, the value
was 1 fatality in 4742 CT exams, which was well within the range computed using
BEIR VII techniques.
11.11 Diagnostic Reference Levels

Now that virtually all radiological imaging procedures are digital, it is possible
to automatically record estimates of the radiation levels used during each examination in radiography, fluoroscopy, mammography and computed tomography
procedures. There are ongoing activities by the American College of Radiology,
National Council on Radiation Protection, and other organizations to develop an
infrastructure which would tally these dose metrics at each institution, for the
purposes of comparing the dose levels between institutions. Dose statistics (such
as mean dose for a chest radiograph, or mean effective dose for abdominal CT)
would be compiled at each institution and compared against national norms. This
exercise permits radiology managers to better understand how well their x-ray
TABLE 11-6
10 y
40 y
70 y
Male risk
0.000254
0.000122
0.0000626
Male (1/risk)
1 in 3940
1 in 8208
1 in 15973
Female Risk
0.000423
0.000186
0.000101
Female (1/risk)
1 in 2364
1 in 5373
1 in 9850
Note: The relative risks, expressed in absolute fractions and as inverted fractions are listed. These data were calculated for a thoracic CT scan, 35 cm in
length, as discussed in the text.
398
TABLE 11-7 ATTRIBUTABLE RISKS FOR CANCER FATALITY

LIFETIME ATTRIBUTABLE RISK OF CANCER MORTALITYa
AGE AT EXPOSURE (YEARS)
Cancer Site
Male patients
Stomach
Colon
10
15
20
30
40
50
60
70
80
41
34
30
25
21
16
15
13
11
163
139
117
99
84
61
60
57
49
36
21
Liver
44
37
31
27
23
16
16
14
12
Lung
318
264
219
182
151
107
107
104
93
71
42
Prostate
17
15
12
10
Bladder
45
38
32
27
23
17
17
17
17
15
10
Other
All solid
Leukemia
All cancers
400
255
200
162
134
94
88
77
58
36
17
1028
781
641
533
444
317
310
289
246
181
102
71
71
71
70
67
64
67
71
73
69
51
1099
852
712
603
511
381
377
360
319
250
153
Female patients
Stomach
57
48
41
34
29
21
20
19
16
13
Colon
86
73
62
53
38
37
35
31
25
15
102
Liver
24
20
17
14
12
Lung
643
534
442
367
305
213
212
204
183
140
81
Breast
274
214
167
130
101
61
35
19
Uterus
11
10
Ovary
55
47
39
34
28
20
20
18
15
10
Bladder
59
51
43
36
31
23
23
22
22
19
13
Other
All solid
Leukemia
All cancers
491
287
220
179
147
103
97
86
69
47
24
1717
1295
1051
862
711
491
455
415
354
265
152
53
52
53
52
51
51
52
54
55
52
38
1770
1347
1104
914
762
542
507
469
409
317
190
Note: Number of deaths per 100,000 persons exposed to a single dose of 0.1 Gy.
a
These estimates are obtained as combined estimates based on relative and absolute risk transport and have
been adjusted by a DDREF of 1.5, except for leukemia, which is based on linear-quadratic model.
Source: Adapted From BEIR VII Table 12D-2.
techniques compare with other institutions, with the notion that institutions producing high dose levels would reduce their dose levels to national norms.
Although recommendations vary, it is thought that institutions who find themselves in the top quartile of doses for a given radiological procedure should implement
dose reduction policies to bring their facility into line with mainstream practice.
As discussed in Chapter 5 (Informatics), the DICOM header contains a great deal
of information and in recent years radiation dose information has been included in
the DICOM standards. This allows the automated extraction of dose data by computer systems which are designed for this. Using this approach, at a busy institution
a large amount of dose data can be acquired in a short period of time, depending
on procedure volume. Although setting up the infrastructure for dose monitoring
requires initial resources as well as ongoing maintenance and analysis, the use of
399
reference dose levels has been embraced both nationally and internationally, and has
already demonstrated its utility in dose reduction.
11.12 Increasing Radiation Burden from Medical Imaging

NCRP 160 recently reported that the average effective dose per person due to medical
imaging in the United States rose from 0.53 mSv in 1987 to 3.0 mSv in 2006. The bulk
of the increased dose from medical procedures was from the increased used of computed tomography, which was thought to contribute about 50% of the ionizing radiation
used for diagnostic imaging. Nuclear medicine procedures accounted for about 25% of
medical radiation. The current average U.S. background radiation level is 3.1 mSv per
year, so medical imaging now contributes about the same as background. Obviously
these numbers represent averages, and many people have no radiological procedures
in a year, while others may have a number of imaging procedures. Table 11-8 describes
the breakdown of the effective doses from various medical imaging procedures.
TABLE 11-8TYPICAL EFFECTIVE DOSES FOR VARIOUS
RADIOGRAPHIC P
ROCEDURES
Procedure
Average Effective Dose (mSv)
Radiographic Procedures
Skull
0.1
Cervical Spine
0.2
Thoracic Spine
1.0
Lumbar Spine
1.5
PA and Lateral Chest
0.1
Mammography
0.4
Abdomen
0.7
Pelvis
0.6
Hip
0.7
Shoulder
0.01
Knee
0.005
Upper GI series
Barium Enemaaa
CT Procedures
Head
Chest
Abdomen
Pelvis
Three phase Liver
15
Spine
CT Colonography
10
iIncludes dose from fluoroscopy.

From Mettler et al. Radiology 2008;248:254263.
400
11.13 Summary: Dose Estimation in Patients

Dose estimation in radiography and fluoroscopy can be performed using a number
of metrics, and comparisons or rule of thumb dose estimates may use ESD as a surrogate indicator of approximate dose levels. In fluoroscopy, the need for dose estimation is often in regard to skin dose, and skin dose estimates can be made relatively
accurately when the exposure conditions and geometry are known. A computation
of radiation dose to an individual organ in the patient from radiography and fluoroscopy requires the use of published tables to convert entrance kerma levels to organ
doses, along with the known exposure conditions and techniques.
There are a number of accurate methods for computing the dose in CT, for the
purposes of technique optimization and for monitoring patient dose levels. In most
cases, the estimation of radiation absorbed dose (in mGy) to a model patient with
dimensions approximate to an actual patient is sufficiently accurate. Dose estimates
for a specific patient can be made by including corrections for patient diameter
(Fig. 11-8) and corrections for the actual scan length used (Fig. 11.11). The figures
shown in this chapter are meant to illustrate concepts, and these data should not be
used for dose calculations on patients. Instead, the readers should refer to root documents and other literature for more specific details on dose estimation in CT.
SUGGESTED READING
Bauhs JA, Vrieze TJ, Primak AN, et al. CT dosimetry: comparison of measurement techniques and devices.
Radiographics 2008;28(1):245253. Review.
Boone JM. The trouble with CTDI100. Med Phys 2007;34(4):13641371.
Boone JM, Buonocore MH, Cooper VN III. Monte Carlo validation in diagnostic radiological imaging. Med
Phys 2000;27:12941304.
Boone JM, Cooper VN, Nemzek WR, et al. Monte Carlo assessment of computed tomography dose to
tissue adjacent to the scanned volume. Med Phys 2000;27:23932407.
DeMarco JJ, Cagnon CH, Cody DD, et al. A Monte Carlo based method to estimate radiation dose
from multidetector CT (MDCT): cylindrical and anthropomorphic phantoms. Phys Med Biol
2005;50(17):39894004.
DeMarco JJ, Cagnon CH, Cody DD, et al. Estimating radiation doses from multidetector CT using Monte
Carlo simulations: effects of different size voxelized patient models on magnitudes of organ and
effective dose. Phys Med Biol 2007;52(9):25832597.
Dixon RL. A new look at CT dose measurement: beyond CTDI. Med Phys 2003;30(6):12721280.
Huda W, Scalzetti EM, Roskopf M. Effective doses to patients undergoing thoracic computed tomography
examinations. Med Phys 2000;27(5):838844.
Jarry G, DeMarco JJ, Beifuss U, et al. A Monte Carlo-based method to estimate radiation dose from spiral
CT: from phantom testing to patient-specific models. Phys Med Biol 2003;48(16):26452663.
Khursheed A, Hillier MC, Shrimpton PC, et al. Influence of patient age on normalized effective doses
calculated for CT examinations. BJR 2002;75:819830.
Linton OW, Mettler FA. National conference on dose reduction in CT, with an emphasis on pediatric
patients. Am J Roentgenol 2003;181:321329.
McCollough CH. CT dose: how to measure, how to reduce. Health Phys 2008;95(5):508517.
McCollough CH, Bruesewitz MR, Kofler JM Jr. CT dose reduction and dose management tools: overview
of available options. Radiographics 2006;26(2):503512. Review.
McCollough CH, Primak AN, Braun N, et al. Strategies for reducing radiation dose in CT. Radiol Clin North
Am 2009;47(1):2740. Review.
McCollough CH, Primak AN, Saba O, et al. Dose performance of a 64-channel dual-source CT scanner.
Radiology 2007;243(3):775784. [Epub 2007 Apr 19.]
McNitt-Gray MF. AAPM/RSNA physics tutorial for residents: topics in CT. Radiation dose in CT.
Radiographics 2002;22(6):15411553. Review.
Menke J. Comparison of different body size parameters for individual dose adaptation in body CT of
adults. Radiology 2005;236:565571.
401
Mettler FA Jr, Wiest PW, Locken JA, et al. CT scanning: patterns of use and dose. J Radiol Prot
2000;20(4):353359.
Shope T, Gagne R, Johnson G. A method for describing the doses delivered by transmission x-ray computed
tomography. Med Phys 1981;8:488495.
Turner AC, Zankl M, DeMarco JJ, et al. The feasibility of a scanner-independent technique to estimate
organ dose from MDCT scans: using CTDIvol to account for differences between scanners. Med Phys
2010;37(4):18161825.
Zhang D, Cagnon CH, Villablanca JP, Cody DD, et. al. Peak Skin and Eye Lens Dose from Brain Perfusion
CT Examinations based on monte carlo simulations, in press. AJR, 2011
Zhou H, Boone JM. Monte Carlo evaluation of CTDI(infinity) in infinitely long cylinders of water,
polyethylene, and PMMA with diameters from 10 mm to 500 mm. Med Phys 2008;35:24242431.
Patient dosimetry for x-rays used in medical imaging. J ICRU Report 74 (2005).
Ionizing radiation exposure to the population of the United States, NCRP Report 160, 2008.
Chapter
12
Magnetic Resonance Basics:

Magnetic Fields, Nuclear
Magnetic Characteristics, Tissue
Contrast, Image Acquisition
Nuclear magnetic resonance (NMR) is the spectroscopic study of the magnetic
properties of the nucleus of the atom. The protons and neutrons of the nucleus
have a magnetic field associated with their nuclear spin and charge distribution.
Resonance is an energy coupling that causes the individual nuclei, when placed
in a strong external magnetic field, to selectively absorb, and later release, energy
unique to those nuclei and their surrounding environment. The detection and
analysis of the NMR signal has been extensively studied since the 1940s as an
analytic tool in chemistry and biochemistry research. NMR is not an imaging technique but rather a method to provide spectroscopic data concerning a sample
placed in a small volume, high field strength magnetic device. In the early 1970s,
it was realized that magnetic field gradients could be used to localize the NMR signal and to generate images that display magnetic properties of the proton, reflecting clinically relevant information, coupled with technological advances and
development of body-size magnets. As clinical imaging applications increased
in the mid-1980s, the nuclear connotation was dropped, and magnetic resonance imaging (MRI), with a plethora of associated acronyms, became commonly
accepted in the medical community.
MR applications continue to expand clinical relevance with higher field strength
magnets, improved anatomic, physiologic, and spectroscopic studies. The high
contrast sensitivity to soft tissue differences and the inherent safety to the patient
resulting from the use of nonionizing radiation have been key reasons why MRI
has supplanted many CT and projection radiography methods. With continuous
improvements in image quality, acquisition methods, and equipment design, MRI is
often the modality of choice to examine anatomic and physiologic properties of the
patient. There are drawbacks, however, including high equipment and siting costs,
scan acquisition complexity, relatively long imaging times, significant image artifacts,
patient claustrophobia, and MR safety concerns.
This chapter reviews the basic properties of magnetism, concepts of resonance,
tissue magnetization and relaxation events, generation of image contrast, and
basic methods of acquiring image data. Advanced pulse sequences, illustration of
image characteristics/artifacts, MR spectroscopy, MR safety, and biologic effects, are
discussed in Chapter 13.
402
Chapter 12 Magnetic Resonance Basics
403
12.1 Magnetism, Magnetic Fields, and Magnets
Magnetism
Magnetism is a fundamental property of matter; it is generated by moving charges,
usually electrons. Magnetic properties of materials result from the organization and
motion of the electrons in either a random or a nonrandom alignment of magnetic
domains, which are the smallest entities of magnetism. Atoms and molecules have
electron orbitals that can be paired (an even number of electrons cancels the magnetic field) or unpaired (the magnetic field is present). Most materials do not exhibit
overt magnetic properties, but one notable exception is the permanent magnet, in
which the individual magnetic domains are aligned in one direction.
Unlike the monopole electric charges from which they are derived, magnetic
fields exist as dipoles, where the north pole is the origin of the magnetic field lines
and the south pole is the return (Fig. 12-1A). One pole cannot exist without the
other. As with electric charges, like magnetic poles repel and opposite poles
attract. Magnetic field strength, B (also called the magnetic flux density), can be conceptualized as the number of magnetic lines of force per unit area, which decreases
roughly as the inverse square of the distance from the source. The SI unit for B is the
Tesla (T). As a benchmark, the earths magnetic field is about 1/20,000 5 0.00005
T 5 0.05 mT. An alternate (historical) unit is the gauss (G), where 1 T 5 10,000 G.
Magnetic Fields
Magnetic fields can be induced by a moving charge in a wire (e.g., see the section on transformers in Chapter 6). The direction of the magnetic field depends
on the sign and the direction of the charge in the wire, as described by the right
hand rule: The fingers point in the direction of the magnetic field when the thumb
points in the direction of a moving positive charge (i.e., opposite to the direction
of electron movement). Wrapping the current-carrying wire many times in a coil
causes a superimposition of the magnetic fields, augmenting the overall strength of
Bar magnet
Current-carrying coiled wire
eN
S
e-
Dipole magnetic field

FIGURE 12-1 A. The magnetic field has two poles with magnetic field lines emerging from the north pole
(N), and returning to the south pole (S), as illustrated by a simple bar magnet. B. A coiled wire carrying an
electric current produces a magnetic field with characteristics similar to a bar magnet. Magnetic field strength
and field density are dependent on the amplitude of the current and the number of coil turns.
404
the magnetic field inside the coil, with a rapid falloff of field strength outside the coil
(see Fig.12-1B). Amplitude of the current in the coil determines the overall magnitude of the magnetic field strength. The magnetic field lines extending beyond the
concentrated field are known as fringe fields.
Magnets
The magnet is the heart of the MR system. For any particular magnet type, performance
criteria include field strength, temporal stability, and field homogeneity. These parameters are affected by the magnet design. Air core magnets are made of wire-wrapped
cylinders of approximately 1-m diameter and greater, over a cylindrical length of 2 to
3 m, where the magnetic field is produced by an electric current in the wires. When
the wires are energized, the magnetic field produced is parallel to the long axis of the
cylinder. In most clinically designed systems, the magnetic field is horizontal and
runs along the cranialcaudal axis of the patient lying supine (Fig. 12-2A). Solid core
magnets are constructed from permanent magnets, a wire-wrapped iron core electromagnet, or a hybrid combination. In these solid core designs, the magnetic field runs
between the poles of the magnet, most often in a vertical direction (Fig. 12-2B). Magnetic fringe fields extend well beyond the volume of the cylinder in air core designs.
Fringe fields are a potential hazard, and are discussed further in Chapter 13.
To achieve a high magnetic field strength (greater than 1 T) requires the electromagnet core wires to be superconductive. Superconductivity is a characteristic of certain metals (e.g., niobiumtitanium alloys) that when maintained at extremely low
temperatures (liquid helium; less than 4K) exhibit no resistance to electric current.
A
Air Core Magnet

X
Y
B0
Fringe Fields Extensive
Solid Core Magnet

Z
X
Y
B0
Fringe Fields Contained

FIGURE 12-2 A. Air core magnets typically have a horizontal main field produced in the bore of the electrical windings, with the z-axis (B0) along the bore axis. Fringe fields for the air core systems are extensive and
are increased for larger bore diameters and higher field strengths. B. The solid core magnet has a vertical field,
produced between the metal poles of a permanent or wire-wrapped electromagnet. Fringe fields are confined
with this design. In both types, the main field is parallel to the z-axis of the Cartesian coordinate system.
405
Superconductivity allows the closed-circuit electromagnet to be energized and

ramped up to the desired current and magnetic field strength by an external electric source. Replenishment of the liquid helium must occur continuously, because if
the temperature rises above a critical value, the loss of superconductivity will occur
and resistance heating of the wires will boil the helium, resulting in a quench.
Superconductive magnets with field strengths of 1.5 to 3 T are common for clinical systems, and 4 to 7 T clinical large bore magnets are currently used for research
applications, with possible future clinical use.
A cross section of the internal superconducting magnet components shows integral parts of the magnet system including the wire coils and cryogenic liquid containment vessel (Fig. 12-3). In addition to the main magnet system, other components
are also necessary. Shim coils interact with the main magnetic field to improve homogeneity (minimal variation of the magnetic flux density) over the volume used for
patient imaging. Radiofrequency (RF) coils exist within the main bore of the magnet
to transmit energy to the patient as well as to receive returning signals. Gradient coils
are contained within the main bore to produce a linear variation of magnetic field
strength across the useful magnet volume.
A magnetic field gradient is obtained by superimposing the magnetic fields
of two or more coils carrying a direct current of specific amplitude and direction
with a precisely defined geometry (Fig. 12-4). The bipolar gradient field varies
over a predefined field of view (FOV), and when superimposed upon B0, a small,
continuous variation in the field strength occurs from the center to the periphery with distance from the center point (the null). Interacting with the much,
much stronger main magnetic field, the subtle linear variations are on the order of
0.005 T/m (5 mT/m) and are essential for localizing signals generated during the
operation of the MR system.
Shim Leads
Instrumentation
Main Leads
Coldhead
Recondenser
Liquid Helium
Superconducting Coils
Thermal
Shield
Helium
Vessel
Active Shims
Shim Coils
Main Coils
Vacuum
Vessel
Gradient
Coils
Specification
Volume
Passive
Shims
RF coils
RF Coils
Imaging
Volume
Main Switch
Vacuum
Insulation
FIGURE 12-3 Internal components of a superconducting air-core magnet are shown. On the left is a cross
section through the long axis of the magnet illustrating relative locations of the components, and on the right
is a simplified cross section across the diameter.
406

Coil
current +
Coil pair
Coil
current -
Magnetic field
Magnetic field
Magnetic field
variation
Distance along coil axis
Superimposed
Gradient: linear change
magnetic fields
in magnetic field
Center of coil pair
FIGURE 12-4 Gradients are produced inside the main magnet with coil pairs. Individual conducting wire
coils are separately energized with currents of opposite direction to produce magnetic fields of opposite
polarity. Magnetic field strength decreases with distance from the center of each coil. When combined, the
magnetic field variations form a linear change between the coils, producing a linear magnetic field gradient,
as shown in the lower graph.
The MR System
The MR system is comprised of several components including those described above,
orchestrated by many processors and control subsystems, as shown in Figure 12-5.
Detail of the individual components, methods of acquiring the MR signals and reconstruction of images are described in the following sections. But first, characteristics
of the magnetic properties of tissues, the resonance phenomenon, and geometric
considerations are explained.
Components
Data
Storage
Shim Power
Supply
Magnet
RF Transmitter
& Receiver
Patient
Table
Digitizer &
Image
Processor
Clock
Gradient
Pulse Program
Gradient
Power Supply
Pulse Program
Measurement
& Control
Host
Computer
Operating
Console
FIGURE 12-5 The MR system is shown (lower left), the operators display (upper left), and the various
subsystems that generate, detect, and capture the MR signals used for imaging and spectroscopy.
407
Magnetic Properties of Materials

Magnetic susceptibility describes the extent to which a material becomes magnetized
when placed in a magnetic field. Induced internal magnetization opposes the external magnetic field and lowers the local magnetic field surrounding the material. On
the other hand, the internal magnetization can form in the same direction as the
applied magnetic field, and increase the local magnetic field. Three categories of susceptibility are defined: diamagnetic, paramagnetic, and ferromagnetic, based upon the
arrangement of electrons in the atomic or molecular structure. Diamagnetic elements
and materials have slightly negative susceptibility and oppose the applied magnetic
field, because of paired electrons in the surrounding electron orbitals. Examples of
diamagnetic materials are calcium, water, and most organic materials (chiefly owing
to the diamagnetic characteristics of carbon and hydrogen). Paramagnetic materials,
with unpaired electrons, have slightly positive susceptibility and enhance the local
magnetic field, but they have no measurable self-magnetism. Examples of paramagnetic materials are molecular oxygen (O2), deoxyhemoglobin, some blood degradation products such as methemoglobin, and gadolinium-based contrast agents. Locally,
these diamagnetic and paramagnetic agents will deplete or augment the local magnetic field (Fig. 12-6), affecting MR images in known, unknown, and sometimes
unexpected ways. Ferromagnetic materials are superparamagneticthat is, they
augment the external magnetic field substantially. These materials, containing iron,
cobalt, and nickel, exhibit self-magnetism in many cases, and can significantly distort the acquired signals.
Magnetic Characteristics of the Nucleus

The nucleus, comprising protons and neutrons with characteristics listed in Table12-1,
exhibits magnetic characteristics on a much smaller scale than for atoms/molecules and
their associated electron distributions. Magnetic properties are influenced by spin and
charge distributions intrinsic to the proton and neutron. A magnetic dipole is created
for the proton, with a positive charge equal to the electron charge but of opposite sign,
due to nuclear spin. Overall, the neutron is electrically uncharged, but subnuclear
charge inhomogeneities and an associated nuclear spin result in a magnetic field of
opposite direction and approximately the same strength as the proton. Magnetic characteristics of the nucleus are described by the nuclear magnetic moment, represented as
Diamagnetic:
Paired electron spins
magnetic field lines
Paramagnetic:
Unpaired electron spins
water molecule
FIGURE 12-6 The local magnetic field can be changed in the presence of diamagnetic (depletion) and paramagnetic (augmentation) materials, with an impact on the signals generated from nearby signal sources such
as the hydrogen atoms in water molecules.
408
TABLE 12-1 PROPERTIES OF THE NEUTRON AND PROTON

CHARACTERISTIC
NEUTRON
PROTON
Mass(kg)
1.674 3 1027
1.672 3 1027
Charge (coulomb)
1.602 3 1019
Spin quantum number
Magnetic moment (J/T)
9.66 3 10
Magnetic moment (nuclear magneton)
1.91
1.41 3 1026
27
2.79
a vector indicating magnitude and direction. For a given nucleus, the nuclear magnetic
moment is determined through the pairing of the constituent protons and neutrons.
If the sum of the number of protons (P) and number of neutrons (N) in the nucleus
is even, the nuclear magnetic moment is essentially zero. However, if N is even and
P is odd, or N is odd and P is even, the resultant noninteger nuclear spin generates a
nuclear magnetic moment. A single nucleus does not generate a large enough nuclear
magnetic moment to be observable, but the conglomeration of large numbers of nuclei
(,1015) arranged in a nonrandom orientation generates an observable nuclear magnetic moment of the sample, from which the MRI signals are derived.
Nuclear Magnetic Characteristics of the Elements

Biologically relevant elements that are candidates for producing MR signals are
listed in Table 12-2. Key features include the strength of the nuclear magnetic
moment, the physiologic concentration, and the isotopic abundance. Hydrogen,
having the largest magnetic moment and greatest abundance, chiefly in water
and fat, is by far the best element for general clinical utility. Other elements
are orders of magnitude less sensitive. Of these, 23Na and 31P have been used
for imaging in limited situations, despite their relatively low sensitivity. Therefore, the nucleus of the hydrogen atom, the proton, is the principal focus for
generating MR signals.
TABLE 12-2MAGNETIC RESONANCE PROPERTIES OF MEDICALLY USEFUL
NUCLEI
SPIN QUANTUM
NUCLEUS NUMBER
% ISOTOPIC
ABUNDANCE
MAGNETIC
MOMENTb
% RELATIVE ELEMENTAL
ABUNDANCEa
RELATIVE
SENSITIVITY
99.98
2.79
10
He
0.00014
22.13
0.011
0.70
18
/2
0.04
21.89
65
9 3 1026
100
2.63
,0.01
3 3 1028
Na
/2
100
2.22
0.1
1 3 1024
100
1.13
1.2
6 3 1025
13
17
19
23
31
moment in nuclear magneton units = 5.05 3 10227 J T21.

Note: by mass in the human body (all isotopes).

A
No magnetic field
409
External magnetic field

Antiparallel spins
Higher energy
DE
B0
Net magnetic
moment
Parallel spins
Lower energy
FIGURE 12-7 Simplified distributions of free protons without and with an external magnetic field are
shown. A. Without an external magnetic field, a group of protons assumes a random orientation of magnetic
moments, producing an overall magnetic moment of zero. B. Under the influence of an applied external magnetic field, B0, the protons assume a nonrandom alignment in two possible orientations: parallel and antiparallel to the applied magnetic field. A slightly greater number of protons exist in the parallel direction, resulting
in a measurable net magnetic moment in the direction of B0.
Magnetic Characteristics of the Proton

The spinning proton or spin (spin and proton are used synonymously herein) is
classically considered to be a tiny bar magnet with north and south poles, even
though the magnetic moment of a single proton is undetectable. Large numbers
of unbound hydrogen atoms in water and fat, those unconstrained by molecular
bonds in complex macromolecules within tissues, have a random orientation of
their protons (nuclear magnetic moments) due to thermal energy. As a result, there
is no observable magnetization of the sample (Fig. 12-7A). However, when placed
in a strong static magnetic field, B0, magnetic forces cause the protons to align with
the applied field in parallel and antiparallel directions at two discrete energy levels
(Fig. 12-7B). Thermal energy within the sample causes the protons to be distributed in this way, and at equilibrium, a slight majority exists in the low-energy,
parallel direction. A stronger magnetic field increases the energy separation of the
low- and high-energy levels and the number of excess protons in the low-energy
state. At 1.0 T, the number of excess protons in the low-energy state is approximately 3 protons per million (3 3 1026) at physiologic temperatures. Although
this number seems insignificant, for a typical voxel volume in MRI, there are
about 1021 protons, so there are 3 3 1026 3 1021, or approximately 3 3 1015
more protons in the low-energy state! This number of excess protons produces an
observable sample nuclear magnetic moment, initially aligned with the direction of the applied magnetic field.
In addition to energy separation of the parallel and antiparallel spin states, the
protons also experience a torque in a perpendicular direction from the applied
magnetic field that causes precession, much the same way that a spinning top wobbles
due to the force of gravity (Fig. 12-8). The precession occurs at an angular frequency
(number of rotations/sec about an axis of rotation) that is proportional to the magnetic field strength B0. The Larmor equation describes the dependence between the
magnetic field, B0, and the angular precessional frequency, 0:
0 5 B0
where g is the gyromagnetic ratio unique to each element. This is expressed in terms
of linear frequency as
410
FIGURE 12-8 A single proton precesses about its axis with an angular
frequency, , proportional to the externally applied magnetic field strength,
according to the Larmor equation. A
well-known example of precession is
the motion a spinning top makes as it
interacts with the force of gravity as
it slows.
w0 radians/s
B0
Gravity
Proton
Spinning top
Precessional frequency
f0 =
B0
2
where 5 2pf and g/2p is the gyromagnetic ratio, with values expressed in millions
of cycles per second (MHz) per Tesla, or MHz/T.
Each element with a nonzero nuclear magnetic moment has a unique gyromagnetic ratio, as listed in Table 12-3.
Energy Absorption and Emission

The protons precessing in the parallel and antiparallel directions result in a
quantized distribution (two discrete energies) with the net magnetic moment of
the sample at equilibrium equal to the vector sum of the individual magnetic
moments in the direction of B0 as shown in Figure 12-9. The magnetic field vector components of the sample in the perpendicular direction are randomly distributed and sum to zero. Briefly irradiating the sample with an electromagnetic
RF energy pulse tuned to the Larmor (resonance) frequency promotes protons
from the low-energy, parallel direction to the higher energy, antiparallel direction, and the magnetization along the direction of the applied magnetic field
shrinks. Subsequently, the more energetic sample returns to equilibrium conditions when the protons revert to the parallel direction and release RF energy at the
TABLE 12-3GYROMAGNETIC RATIO FOR USEFUL
ELEMENTS IN MAGNETIC RESONANCE

NUCLEUS
1
g/2p (MHz/T)
42.58
10.7
5.8
40.0
Na
11.3
17.2
13
17
19
23
31
M0
Parallel
B0
AntiParallel
411
FIGURE 12-9 A group of protons

in the parallel and antiparallel energy
states generates an equilibrium magnetization, M0, in the direction of the
applied magnetic field B0. The protons are distributed randomly over
the surface of the cone, and produce
no magnetization in the perpendicular direction.
Group of protons net

magnetized sample
same frequency. This energy emission is detected by highly sensitive antennas to

capture the basic MRsignal.
Typical magnetic field strengths for MR systems range from 0.3 to 4.0 T. For protons, the precessional frequency is 42.58 MHz/T, and increases or decreases with an
increase or decrease in magnetic field strength, as calculated in the example below.
Accuracy of the precessional frequency is necessary to ensure that the RF energy
will be absorbed by the magnetized protons. Precision of the precessional frequency
must be on the order of cycles/s (Hz) out of millions of cycles/s (MHz) in order to
identify the location and spatial position of the emerging signals, as is described in
Section 12.6.
Example: What is the frequency of precession of 1H and 31P at 0.5 T? 1.5 T? 3.0 T?
The Larmor frequency is calculated as f0 = (/2)B0.
Field Strength
Element
1
31
0.5 T
1.5 T
3.0 T
42.58 3 0.5 = 21.29 MHz 42.58 3 1.5 = 63.87 MHz 42.58 3 3.0 = 127.74 MHz
17.2 3 0.5 = 8.6 MHz
17.2 3 1.5 = 25.8 MHz
17.2 3 3 = 51.6 MHz
The differences in the gyromagnetic ratios and corresponding precessional frequencies allow the selective excitation of one element from another in the same magnetic
field strength.
Geometric Orientation, Frame of Reference, and

Magnetization Vectors
By convention, the applied magnetic field B0 is directed parallel to the z-axis of the
three-dimensional Cartesian coordinate axis system and perpendicular to the x and y
axes. For convenience, two frames of reference are used: the laboratory frame and the
rotating frame. The laboratory frame (Fig. 12-10A) is a stationary reference frame from
the observers point of view. The sample magnetic moment vector precesses about the
412

A
Laboratory Frame
z
Rotating Frame
z
B0
Precessing
moment is
stationary
y'
x'
x y axes stationary
x y axes rotate at Larmor frequency
FIGURE 12-10 A. The laboratory frame of reference uses stationary three-dimensional Cartesian coordinates: x, y, z. The magnetic moment precesses around the z-axis at the Larmor frequency as the illustration
attempts to convey. B. The rotating frame of reference uses rotating Cartesian coordinate axes that rotate
about the z-axis at the Larmor precessional frequency, and the other axes are denoted: x and y. When precessing at the Larmor frequency, the sample magnetic moment is stationary.
z-axis in a circular geometry about the x-y plane. The rotating frame (Fig. 12-10B) is
a spinning axis system, whereby the x-y axes rotate at an angular frequency equal to
the Larmor frequency. In this frame, the sample magnetic moment vector appears to
be stationary when rotating at the resonance frequency. A slightly higher precessional
frequency is observed as a slow clockwise rotation, while a slightly lower precessional
frequency is observed as a slow counterclockwise rotation. The magnetic interactions
between precessional frequencies of the tissue magnetic moments with the externally
applied RF (depicted as a rotating magnetic field) can be described more clearly using
the rotating frame of reference, while the observed returning signal and its frequency
content is explained using the laboratory (stationary) frame of reference.
The net magnetization vector of the sample, M, is described by three components. Longitudinal magnetization, Mz, along the z direction, is the component of the
magnetic moment parallel to the applied magnetic field, B0. At equilibrium, the longitudinal magnetization is maximal and is denoted as M0, the equilibrium magnetization. The component of the magnetic moment perpendicular to B0, Mxy, in the x-y
plane, is transverse magnetization. At equilibrium, Mxy is zero. When the protons in
the magnetized sample absorb energy, Mz is tipped into the transverse plane, and
Mxy generates the all-important MR signal. Figure 12-11 illustrates this geometry.
12.2 The Magnetic Resonance Signal

Application of RF energy synchronized to the precessional frequency of the protons
causes absorption of energy and displacement of the sample magnetic moment from
equilibrium conditions. The return to equilibrium results in the emission of energy
proportional to the number of excited protons in the volume. This occurs at a rate
that depends on the structural and magnetic characteristics of the sample. Excitation,
detection, and acquisition of the signals constitute the basic information necessary
for MRI and and MR spectroscopy (MRS).
413
M0
B0
Mz
y
Mxy
x
Mz: Longitudina l Magnetization: in z-axis direction
Mxy: Transverse Magnetization:
in x-y plane
M0: Equilibrium Magnetization:
maximum magnetization
FIGURE 12-11 Longitudinal magnetization, Mz, is the vector component of the magnetic moment in the z
direction. Transverse magnetization, Mxy, is the vector component of the magnetic moment in the x-y plane.
Equilibrium magnetization, M0, is the maximal longitudinal magnetization of the sample, and is shown displaced from the z-axis in this illustration.
Resonance and Excitation

Displacement of the equilibrium magnetization occurs when the magnetic component of the RF excitation pulse, known as the B1 field, is precisely matched to the
precessional frequency of the protons. The resonance frequency corresponds to the
energy separation between the protons in the parallel and antiparallel directions.
The quantum mechanics model considers the RF energy as photons (quanta)
instead of waves. Protons oriented parallel and antiparallel to the external magnetic
field, separated by an energy gap, DE, will transition from the low- to the highenergy level only when the RF pulse is equal to the precessional frequency. The
number of protons that undergo an energy transition is dependent on the amplitude and duration of the RF pulse. Mz changes from the maximal positive value at
equilibrium, through zero, to the maximal negative value (Fig. 12-12). Continued
z
Equilibrium
more spins
parallel than
antiparallel
RF energy (B 1)
applied to the
system at the
Larmor Frequenc y
Mz positive
Equal numbers
of parallel and
antiparallel spins
Time of B 1 field
increasing
Mz = 0
B0
Excited spins
occupy antiparallel energy
levels
More antiparallel
than parallel
Mz negative
FIGURE 12-12 A simple quantum mechanics process depicts the discrete energy absorption and the time
change of the longitudinal magnetization vector as RF energy equal to the energy difference of the parallel
and antiparallel spins is applied to the sample (at the Larmor frequency). Discrete quanta absorption changes
the proton energy from parallel to antiparallel. With continued application of the RF energy at the Larmor
frequency, Mz is displaced from equilibrium, through zero, to the opposite direction (high energy state).

A
Magnetic field variation of electromagnetic RF wave
Amplitude
414
Clockwise Counterrotating clockwise

vector
rotating
vector
Time
B1 at Larmor Frequency
z
Mz
B1
y'
B1
x'
B 1 off resonance
z
Direction of
torque on Mz
Mz
B1
B1
y'
B1
x'
B1
B1
B1
FIGURE 12-13 A. A classical physics description of the magnetic field component of the RF pulse (the
electric field is not shown). Clockwise (solid) and counterclockwise (dotted) rotating magnetic vectors produce
the magnetic field variation by constructive and destructive interaction. At the Larmor frequency, one of the
magnetic field vectors rotates synchronously in the rotating frame and is therefore stationary (the other vector
rotates in the opposite direction and does not synchronize with the rotating frame). B. In the rotating frame,
the RF pulse (B1 field) is applied at the Larmor frequency and is stationary in the x-y plane. The B1 field interacts at 90 degrees to the sample magnetic moment and produces a torque that displaces the magnetic vector
away from equilibrium. C. The B1 field is not tuned to the Larmor frequency and is not stationary in the rotating
frame. No interaction with the sample magnetic moment occurs.
irradiation can induce a return to equilibrium conditions, when an incoming RF

quantum of energy causes the reversion of a proton in the antiparallel direction
to the parallel direction with the energy-conserving spontaneous emission of two
excess quanta. While this model shows how energy absorption and emission occurs,
there is no clear description of how Mxy evolves, which is better understood using
classical physics concepts.
In the classical physics model, the linear B1 field is described with two magnetic field vectors of equal magnitude, rotating in opposite directions, representing
the sinusoidal variation of the magnetic component of the electromagnetic RF wave
as shown in Figure 12-13. One of the two rotating vectors is synchronized to the
precessing protons in the magnetized sample, and in the rotating frame is stationary. If the RF energy is not applied at the precessional (Larmor) frequency, the B1
field will not interact with Mz. Another description is that of a circularly polarized
B1 transmit field from the body coil that rotates at the precessional frequency of the
magnetization.
Flip Angles
Flip angles represent the degree of Mz rotation by the B1 field as it is applied along
the x-axis (or the y-axis) perpendicular to Mz. A torque is applied on Mz, rotating it
from the longitudinal direction into the transverse plane. The rate of rotation occurs

A Small flip angle
B Large flip angle
DMz
M0
DMz
Mz
Mxy
B1
M0
Mz
y'
Mxy
B1
x'
y'
x'
z
C Common flip angles
B1
180 flip
90 flip
x'
415
Mxy
y'
B1
x'
y'
-Mz
FIGURE 12-14 Flip angles are the result of the angular displacement of the longitudinal magnetization
vector from the equilibrium position. The rotation angle of the magnetic moment vector is dependent on the
duration and amplitude of the B1 field at the Larmor frequency. Flip angles describe the rotation of Mz away
from the z-axis. A. Small flip angles (less than 45 degrees) and B. Large flip angles (75 to 90 degrees) produce
small and large transverse magnetization, respectively.
at an angular frequency equal to 1 5 gB1, as per the Larmor equation. Thus, for
an RF pulse (B1 field) applied over a time t, the magnetization vector displacement
angle, , is determined as 5 1 t 5 gB1t, and the product of the pulse time and B1
amplitude determines the displacement of Mz. This is illustrated in Figure 12-14.
Common flip angles are 90 degrees (p/2) and 180 degrees (p), although a variety of
smaller and larger angles are chosen to enhance tissue contrast in various ways. A 90-degree angle provides the largest possible Mxy and detectable MR signal, and requires a
known B1 strength and time (on the order of a few to hundreds of ms). The displacement
angle of the sample magnetic moment is linearly related to the product of B1 field strength
and time: For a fixed B1 field strength, a 90-degree displacement takes half the time of a
180-degree displacement. With flip angles smaller than 90 degrees, less time is needed
to displace Mz, and a larger transverse magnetization per unit excitation time is achieved.
For instance, a 45-degree flip takes half the time of a 90 degrees yet creates 70% of the
signal, as the magnitude of Mxy is equal to the sine of 45 degrees, or 0.707. With fast MRI
techniques, small displacement angles of 10 degrees and less are often used.
12.3 Magnetization Properties of Tissues
Free Induction Decay: T2 Relaxation

After a 90-degree RF pulse is applied to a magnetized sample at the Larmor frequency,
an initial phase coherence of the individual protons is established and maximum Mxy
is achieved. Rotating at the Larmor frequency, the transverse magnetic field of the
excited sample induces signal in the receiver antenna coil (in the laboratory frame of
reference). A damped sinusoidal electronic signal, known as the free induction decay
(FID), is produced (Fig. 12-15).
The FID amplitude decay is caused by loss of Mxy phase coherence due to intrinsic
micromagnetic inhomogeneities in the samples structure, whereby individual protons
416

z
Rotating frame
y'
x'
Equilibrium
Mxy = zero
90 RF pulse
Mxy large
Dephasing
Mxy decreasing
Dephased
Mxy = zero
Time
Laboratory frame
z
Rotating Mxy vector

induces signal in antenna
+
90
y
y
x
FID
Time
FIGURE 12-15 Top. Conversion of longitudinal magnetization, Mz, into transverse magnetization, Mxy ,
results in an initial phase coherence of the individual spins of the sample. The magnetic moment vector precesses at the Larmor frequency (stationary in the rotating frame), and dephases with time. Bottom. In the
laboratory frame, Mxy precesses and induces a signal in an antenna receiver sensitive to transverse magnetization. A FID signal is produced with positive and negative variations oscillating at the Larmor frequency, and
decaying with time due to the loss of phase coherence.
in the bulk water and hydration layer coupled to macromolecules precess at incrementally different frequencies arising from the slight changes in local magnetic field strength.
Phase coherence is lost over time as an exponential decay. Elapsed time between the
peak transverse signal (e.g., directly after a 90-degree RF pulse) and 37% of the peak
level (1/e) is the T2 relaxation time (Fig. 12-16A). Mathematically, this is expressed as
Mxy (t) 5 M0e2t/T2
where Mxy(t) is the transverse magnetic moment at time t for a sample that
has M0 transverse magnetization at t 5 0. When t 5 T2, then e21 5 0.37 and
Mxy 5 0.37 M0.
The molecular structure of the magnetized sample and characteristics of the
bound water protons strongly affects its T2 decay value. Amorphous structures (e.g.,
cerebral spinal fluid [CSF] or highly edematous tissues) contain mobile molecules
with fast and rapid molecular motion. Without structural constraint (e.g., lack of
a hydration layer), these tissues do not support intrinsic magnetic field inhomogeneities, and thus exhibit long T2 values. As molecular size increases for specific
tissues, constrained molecular motion and the presence of the hydration layer produce magnetic field domains within the structure and increase spin dephasing that
causes more rapid decay with the result of shorter T2 values. For large, nonmoving
structures, stationary magnetic inhomogeneities in the hydration layer result in these
types of tissues (e.g., bone) having a very short T2.
Extrinsic magnetic inhomogeneities, such as the imperfect main magnetic field,
B0, or susceptibility agents in the tissues (e.g., MR contrast materials, paramagnetic or
ferromagnetic objects), add to the loss of phase coherence from intrinsic inhomogeneities and further reduce the decay constant, known as T2* under these conditions
(Fig. 12-16B).

A
417
B
Mxy
maximum
Mxy
decreasing
Mxy
zero
Mxy
Mxy
100%
T2 decay
T2* decay
37%
Time
t=0 t=T2
Time
FIGURE 12-16 A. The loss of Mxy phase coherence occurs exponentially caused by intrinsic spin-spin interactions in the tissues and extrinsic magnetic field inhomogeneities. The exponential decay constant, T2, is
the time over which the signal decays to 37% of the initial transverse magnetization (e.g., after a 90-degree
pulse). B. T2 is the decay time resulting from intrinsic magnetic properties of the sample. T2* is the decay time
resulting from both intrinsic and extrinsic magnetic field variations. T2 is always longer than T2*.
Return to Equilibrium: T1 Relaxation

Longitudinal magnetization begins to recover immediately after the B1 excitation
pulse, simultaneous with transverse decay; however, the return to equilibrium conditions occurs over a longer time period. Spin-lattice relaxation is the term describing
the release of energy back to the lattice (the molecular arrangement and structure of
the hydration layer), and the regrowth of Mz. This occurs exponentially as
Mz (t) 5 M0(1 2 et/T1)
where Mz(t) is the longitudinal magnetization at time t and T1 is the time needed for the
recovery of 63% of Mz after a 90-degree pulse (at t 5 0, Mz 5 0, and at t 5 T1, Mz 5
0.63 M0), as shown in Figure 12-17. When t 5 3 3 T1, then Mz 5 0.95 M0, and for t .
5 3 T1, then Mz < M0, and full longitudinal magnetization equilibrium is established.
Since Mz does not generate an MR signal directly, determination of T1 for a specific tissue requires a specific sequence, as shown in Figure 12-18. At equilibrium,
a 90-degree pulse sets Mz 5 0. After a delay time, DT, the recovered Mz component
is converted to Mxy by a second 90-degree pulse, and the resulting peak amplitude
Mz
Mz
Mxy
90
pulse
100%
63%
Mz
0%
t=0
t = T1
Time
FIGURE 12-17 After a 90-degree pulse, Mz is converted from a maximum value at equilibrium to Mz = 0.
Return of Mz to equilibrium occurs exponentially and is characterized by the spin-lattice T1 relaxation constant.
After an elapsed time equal to T1, 63% of the longitudinal magnetization is recovered. Spin-lattice recovery
takes longer than spin-spin decay (T2).
418
FIGURE 12-18 Spin-lattice relaxation for a sample can be measured by using various delay times
between two 90-degree RF pulses. After an initial
90-degree pulse, Mz = 0, another 90-degree pulse
separated by a known delay is applied, and the
longitudinal magnetization that has recovered during the delay is converted to transverse magnetization. The maximum amplitude of the resultant FID
is recorded as a function of delay times between
initial pulse and readout (three different delay time
experiments are shown in this example), and the
points are fit to an exponential recovery function
to determine T1.
Equilibrium
x
90 pulse
x
Delay time
Longitudinal
recovery
100% Mz
0% Mz
short
y
x
z
90 pulse
(readout)
medium
long
z
y
Resulting FID
100%
Mz recovery
0%
Delay time (s)
is recorded. By repeating the sequence from equilibrium conditions with different

delay times, DT between 90-degree pulses, data points that lie on the recovery curve
are fit to an exponential equation and T1 is estimated.
The T1 relaxation time depends on the rate of energy dissipation into the surrounding molecular lattice and hydration layer and varies substantially for different
tissue structures and pathologies. This can be explained from a classical physics
perspective by considering the tumbling frequencies of the protons in bulk water
and the hydration layers present relative to the Larmor precessional frequency of the
protons. Energy transfer is most efficient when a maximal overlap of these frequencies occurs. Small molecules and unbound, bulk water have tumbling frequencies
across a broad spectrum, with low-, intermediate-, and high-frequency components.
Large, slowly moving molecules have a very tight hydration layer and exhibit low
tumbling frequencies that concentrate in the lowest part of the frequency spectrum.
Moderately sized molecules (e.g., proteins) and viscous fluids have a moderately
bound hydration layer that produce molecular tumbling frequencies more closely
matched to the Larmor frequency. This is more fully described in the Two Compartment Fast Exchange Model (Fullerton, et.al, 1982; Bottomley, et. al, 1984).
The water in the hydration layer and bulk water exchange rapidly (on the order of
100,000 transitions per second) so that a weighted average of the T1 is observed.
Therefore, the T1 time is strongly dependent on the physical characteristics of the
tissues and their associated hydration layers. Therefore for solid and slowly moving
structures, the hydration layer permits only low-frequency molecular tumbling frequencies and consequently, there is almost no spectral overlap with the Larmor frequency. For unstructured tissues and fluids in bulk water, there is also only a small
spectral overlap with the tumbling frequencies. In each of these situations, release of
energy is constrained and T1 relaxation time is long. For structured and moderately
sized proteins and fatty tissues, molecular tumbling frequencies are most conducive
to spin-lattice relaxation because of a larger overlap with the Larmor frequency and
result in a relatively short T1 relaxation time, as shown in Figure 12-19. Typical T1
values are in the range of 0.1 to 1 s for soft tissues, and 1 to 4 s in aqueous tissues
(e.g., CSF).
As the main magnetic field strength increases, a corresponding increase in the
Larmor precessional frequency causes a decrease in the overlap with the molecular tumbling frequencies and a longer T1 recovery time. Gadolinium chelated with
w0
B0
Higher B0
419
w0
Relative Amplitude
Medium, viscous:
Short T1
Small, aqueous:
Long T1
Large, stationary:
Longest T1
Low
Frequency
High
Molecular vibration frequency spectrum

FIGURE 12-19 A classical physics explanation of spin-lattice relaxation is based upon the tumbling frequency
of the protons in water molecules of a sample material and its frequency spectrum. Large, stationary structures have water protons with tight hydration layers that exhibit little tumbling motion and a low-frequency
spectrum (aqua curve). Bulk water and small-sized, aqueous materials have frequencies distributed over a
broad range (purple curve). Medium-sized, proteinacious materials have a hydration layer that slows down
the tumbling frequency of protons sufficiently to allow tumbling at the Larmor frequency (black curve). The
overlap of the Larmor precessional frequency (vertical bar) with the molecular vibration spectrum indicates
the likelihood of spin-lattice relaxation. With higher field strengths, the T1 relaxation becomes longer as the
tumbling frequency spectrum overlap is decreased.
c omplex macromolecules are effective in decreasing T1 relaxation time of local

tissues by creating a hydration layer that forms a spin-lattice energy sink and results
in a rapid return to equilibrium.
Comparison of T1 and T2
T1 is on the order of 5 to 10 times longer than T2. Molecular motion, size, and
interactions influence T1 and T2 relaxation (Fig. 12-20). Because most tissues of
interest for clinical MR applications are intermediate to small-sized molecules, tissues
Long
T1
T2
Relaxation
time
Short
Molecular motion: slow
intermediate
fast
Molecular size: large
intermediate
small
intermediate
free
Molecular interactions: bound
FIGURE 12-20 Factors affecting T1 and T2 relaxation times of different tissues are generally based on
molecular motion, size, and interactions that have an impact on the local magnetic field variations (T2 decay)
and structure with intrinsic tumbling frequencies coupling to the Larmor frequency (T1 recovery). The relaxation times (vertical axis) are different for T1 and T2.
420
TABLE 12-4 T1 AND T2 RELAXATION CONSTANTS FOR SEVERAL TISSUESa

TISSUE
T1, 0.5 T (ms)
T1, 1.5 T (ms)
T2 (ms)
Fat
210
260
80
Liver
350
500
40
Muscle
550
870
45
White matter
500
780
90
Gray matter
650
900
100
1,800
2,400
160
Cerebrospinal fluid
a
Estimates only, as reported values for T1 and T2 span a wide range.
with a longer T1 usually have a longer T2, and those with a shorter T1 usually have
a shorter T2. In Table 12-4, a comparison of T1 and T2 values for various tissues
is listed. Depending on the main magnetic field strength, measurement methods,
and biological variation, these relaxation values vary widely. Agents that disrupt the
local magnetic field environment, such as paramagnetic blood degradation products, elements with unpaired electron spins (e.g., gadolinium), or any ferromagnetic
materials, cause a significant decrease in T2*. In situations where a macromolecule
binds free water into a hydration layer, T1 is also significantly decreased.
To summarize, T1 . T2 . T2*, and the specific relaxation times are a characteristic of the tissues. T1 values are longer for higher field strength magnets, while T2
values are unaffected. Thus, the T1, T2, and T2* decay constants, as well as proton
density are fundamental properties of tissues, and can be exploited by machine-dependent acquisition techniques in MRI and MRS to aid in the diagnosis of pathologic
conditions such as cancer, multiple sclerosis, or hematoma.
12.4 Basic Acquisition Parameters

Emphasizing the differences of T1 and T2, relaxation time constants, and proton
density of the tissues is the key to the exquisite contrast sensitivity of MR images, but
at the same time, the need to spatially localize the tissues is also required. First, basic
machine-based parameters are described.
Time of Repetition
Acquiring an MR image relies on the repetition of a sequence of events in order to
sample the volume of interest and periodically build the complete dataset over time.
The time of repetition (TR) is the period between B1 excitation pulses. During the
TR interval, T2 decay and T1 recovery occur in the tissues. TR values range from
extremely short (millisecond) to extremely long (10,000 ms) time periods, determined by the type of sequence employed.
Time of Echo
Excitation of protons with the B1 RF pulse creates the Mxy FID signal. To separate
the RF energy deposition and returning signal, an echo is induced to appear at a
later time, with the application of a 180-degree RF inversion pulse. This can also be
achieved with a gradient field and subsequent polarity reversal. The time of echo (TE)
is the time between the excitation pulse and the appearance of the peak amplitude of
421
an induced echo, which is determined by applying a 180-degree RF inversion pulse

or gradient polarity reversal at a time equal to TE/2.
Time of Inversion
The TI is the time between an initial inversion/excitation (180 degrees) RF pulse that
produces maximum tissue saturation, and a 90-degree readout pulse. During the TI,
Mz recovery occurs. The readout pulse converts the recovered Mz into Mxy, which is
then measured with the formation of an echo at time TE as discussed above.
Partial Saturation
Saturation is a state of tissue magnetization from equilibrium conditions. At equilibrium, the protons in a material are unsaturated, with full Mz amplitude. The first
excitation (B1) pulse in the sequence produces the largest transverse magnetization,
and recovery of the longitudinal magnetization occurs at the T1 time constant over
the TR interval. However, because the TR is less than at least five times the T1 of the
sample, Mz recovery is incomplete. Consequently, less Mxy amplitude is generated
in the second excitation pulse. After the third pulse, a steady-state equilibrium is
reached, where the amount of Mz recovery and Mxy signal amplitude are constant,
and the tissues achieve a state of partial saturation (Fig. 12-21). Tissues with short
T1 have relatively less saturation than tissues with long T1. Partial saturation has an
impact on tissue contrast, and explains certain findings such as unsaturated protons
in blood outside of the volume moving into the volume and generating a bright
vascular signal on entry slices into the volume.
12.5 Basic Pulse Sequences

Three major pulse sequences perform the bulk of data acquisition (DAQ) for imaging:
spin echo (SE), inversion recovery (IR), and gradient echo (GE). When used in conjunction with spatial localization methods, contrast-weighted images are obtained.
In the following sections, the salient points and considerations of generating tissue
contrast are discussed.
Mz
unsaturated
unsaturated
partially
saturated
Short T1
Long T1
partially
saturated
TR
90
TR
90
TR
90
TR
TR
90
90
90
FIGURE 12-21 Partial saturation of tissues occurs because the repetition time between excitation pulses does
not allow for full return to equilibrium, and the Mz amplitude for the next RF pulse is reduced. After the third
excitation pulse, a steady-state equilibrium is reached, where the amount of longitudinal magnetization is the
same from pulse to pulse, as is the transverse magnetization for a tissue with a specific T1 decay constant. Tissues
with long T1 experience a greater partial saturation than do tissues with short T1 as shown above. Partial saturation is important in understanding contrast mechanisms and signal from unsaturated and saturated tissues.
422
Spin Echo
SE describes the excitation of the magnetized protons in a sample with a 90-degree
RF pulse and production of a FID, followed by a refocusing 180-degree RF pulse to
produce an echo. The 90-degree pulse converts Mz into Mxy , and creates the largest
phase coherent transverse magnetization that immediately begins to decay at a rate
described by T2* relaxation. The 180-degree RF pulse, applied at TE/2, inverts the
spin system and induces phase coherence at TE, as depicted in the rotating frame in
Figure 12-22. Inversion of the spin system causes the protons to experience external
magnetic field variations opposite of that prior to TE/2, resulting in the cancellation
of the extrinsic inhomogeneities and associated dephasing effects. In the rotating
frame of reference, the echo magnetization vector reforms in the opposite direction
from the initial transverse magnetization vector.
Subsequent 180-degree RF pulses during the TR interval (Fig. 12-23) produce
corresponding echoes with peak amplitudes that are reduced by intrinsic T2 decay
of the tissues, and are immune from extrinsic inhomogeneities. Digital sampling and
acquisition of the signal occurs in a time window symmetric about TE, during the
evolution and decay of each echo.
Spin Echo Contrast Weighting

Contrast is proportional to the difference in signal intensity between adjacent pixels
in an image, corresponding to different voxels in the patient. The details of signal
localization and image acquisition in MRI are discussed in Section 12.6. Here, the
signal intensity variations for different tissues based upon TR and TE settings are
described without consideration of spatial localization.
Ignoring the signal due to moving protons (e.g., blood flow), the signal intensity
produced by an MR system for a specific tissue using a SE sequence is
S rH [1 2 eTR/T1] eTE/T2
Excitation
Echo
TE
90
180
TE / 2
Rotating frame
M xy
After 180 pulse, echo

reforms at same rate
FID signal gradually decays
with rate constant T2*
Spin echo peak amplitude

depends on T2
FIGURE 12-22 The SE pulse sequence starts with a 90-degree pulse and produces an FID that decays
according to T2* relaxation. After a delay time TE/2, a 180-degree RF pulse inverts the spins that re-establishes
phase coherence and produces an echo at a time TE. Inhomogeneities of external magnetic fields are canceled,
and the peak amplitude of the echo is determined by T2 decay. The rotating frame shows the evolution of the
echo vector in the opposite direction of the FID. The sequence is repeated for each repetition period, TR.
90
pulse
180
pulse
180
pulse
423
180
pulse
T2 decay
T2* decay
FIGURE 12-23 True T2 decay is determined from multiple 180-degree refocusing pulses acquired during
the repetition period. While the FID envelope decays with the T2* decay constant, the peak amplitudes of
subsequent echoes decay exponentially according to the T2 decay constant, as extrinsic magnetic field inhomogeneities are cancelled.
where rH is the proton density, T1 and T2 are physical properties of tissue, and TR
and TE are pulse sequence timing parameters. For the same pulse sequence, different
values of T1, T2, and rH change the signal intensity S, and generate contrast amongst
different tissues. Importantly, by changing the pulse sequence parameters TR and TE,
the contrast dependence can be weighted toward T1, proton density, or T2 characteristics of the tissues.
T1 Weighting
A T1-weighted SE sequence is designed to produce contrast chiefly based on the T1
characteristics of tissues, with de-emphasis of T2 and proton density contributions to
the signal. This is achieved by using a relatively short TR to maximize the differences
in longitudinal magnetization recovery during the return to equilibrium, and a short
TE to minimize T2 decay during signal acquisition. In Figure 12-24, on the left is the
graph of longitudinal recovery in steady-state partial saturation after a 90-degree RF
excitation at time t 5 0, depicting four tissues (CSF, gray matter, white matter, and
fat). The next 90-degree RF pulse occurs at the selected TR interval, chosen to create
the largest signal difference between the tissues based upon their respective T1 recovery values, which is shown to be about 600 ms (the red vertical line). At this instant
in time, all Mz recovered for each tissue is converted to Mxy, with respective signal
amplitudes projected over to the transverse magnetization graph on the right. Decay
immediately occurs, t 5 0, at a rate based upon respective T2 values of the tissues.
To minimize T2 decay and to maintain the differences in signal amplitude due to T1
recovery, the TE time is kept short (red vertical line). Horizontal projections from
the TE intersection with each of the curves graphically illustrate the relative signal
amplitudes acquired according to tissue type. Fat, with a short T1, has a large signal,
because there is greater recovery of the Mz vector over the TR period. White and gray
matter have intermediate T1 values with intermediate signal amplitude, and CSF,
with a long T1, has the lowest signal amplitude. A short TE preserves the T1 signal
differences by not allowing any significant transverse (T2) decay.
T1-weighted SE contrast therefore requires a short TR and a short TE.
AT1-weighted axial image of the brain acquired with TR 5 500 ms and TE 5 8 ms
is illustrated in Figure 12-25. Fat is the most intense signal, followed by white matter,
gray matter, and CSF. Typical SE T1-weighting machine parameters are TR 5 400 to
600 ms and TE 5 3 to 10 ms.
424
Relative Signal Recovery
Longitudinal recovery (T1)

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Transverse decay (T2)
Fat
Fat
White
Gray
White
Gray
CSF
TR
1000 2000 3000 4000 5000 0

Time (ms)
TE
Signal
intensity
CSF
100
200
300
400
500
Time (ms)
FIGURE 12-24 T1-weighted contrast: Longitudinal recovery (left) and transverse decay (right) diagrams
(note the values of the x-axis time scales) show four brain tissues and T1 and T2 relaxation constants.
T1-weighted contrast requires the selection of a TR that emphasizes the differences in the T1 characteristics
of the tissues (e.g., TR = , 500 ms), and reduces the T2 characteristics by using a short TE so that transverse
decay is reduced (e.g., TE 15 ms).
Proton Density Weighting

Proton density contrast weighting relies mainly on differences in the number of magnetized protons per unit volume of tissue. At equilibrium, tissues with a large proton density, such as lipids, fats, and CSF, have a corresponding large Mz compared
to other soft tissues. Contrast based on proton density differences is achieved by
reducing the contributions of T1 recovery and T2 decay. T1 differences are reduced
by selecting a long TR value to allow substantial recovery of Mz. T2 differences of
the tissues are reduced by selecting a short TE value. Longitudinal recovery and
transverse decay graphs for proton density weighting, using a long TR and a short
TE, are illustrated in Figure 12-26. Contrast is generated from variations in proton
density (CSF . fat . gray matter . white matter). Figure 12-27 shows a proton
FIGURE 12-25 T1 contrast weighting, TR=500 ms,

TE=8ms. Short TR (400 to 600 ms) generates T1 relaxationdependent signals. Signals with short T1 have high signal
intensity (fat and white matter), while signals with long T1
have low signal intensity (CSF). Short TE (less than 15 ms)
preserves the T1 tissue differences by not allowing significant T2 decay to occur.
Relative Signal Intensity

1
0.9
0.8 Fat
White
0.7
Gray
0.6
CSF
0.5
0.4
0.3
0.2
0.1
0
TR
425
Signal
intensity
CSF
Gray
Fat
White
1000 2000 3000 4000 5000 0

Time (ms)
TE
100
200
300
400
500
Time (ms)
FIGURE 12-26 Proton density weighting: Proton (spin) density weighted contrast requires the use of a long
TR (e.g., greater than 2,000 ms) to reduce T1 effects, and a short TE (e.g., less than 35 ms) to reduce T2 influence in the acquired signals. Note that the average overall signal intensity is higher.
density-weighted image with TR 5 2,400 ms and TE 5 30 ms. Fat and CSF display
as a relatively bright signal, and a slight contrast inversion between white and gray
matter occurs. Atypical proton density-weighted image has a TR between 2,000 and
4,000 ms (see footnote in Table 12.5) and a TE between 3 and 30 ms. The proton
density SE sequence achieves the highest overall signal intensity and the largest signal-to-noise ratio (SNR); however, the image contrast is relatively low, and therefore
the contrast-to-noise ratio is not necessarily larger than achievable with T1 or T2
contrast weighting.
T2 Weighting
T2 contrast weighting follows directly from the proton density-weighting sequence:
reduce T1 differences in tissues with a long TR, and emphasize T2 differences with
a long TE. The T2-weighted signal is generated from the second echo produced by a
FIGURE 12-27 Proton density contrast weighting,

TR=2400 ms, TE=30 ms.. Long TR minimizes T1 relaxation
differences of the tissues. Signals with large proton density have higher signal intensity (CSF). Short TE preserves
the proton density differences without allowing significant
T2 decay. This sequence produces a high peak SNR, even
though the contrast differences are less than a T2-weighted
image.
426
Relative Signal Intensity
1
0.9
0.8 FatWhite
0.7
Gray
0.6
CSF
0.5
0.4
0.3
0.2
0.1
0
0 1000 2000 3000 4000 5000 0
TR
Time (ms)
TE
Signal
intensity
CSF
Gray
White
Fat
100
200
300
400
500
Time (ms)
FIGURE 12-28 T2 weighted contrast requires the use of a long TR (e.g., greater than 2,000 ms) to reduce
T1 influences, and a long TE (e.g., greater than 80 ms) to allow for T2 decay to evolve. Compared to the proton
density weighting, the difference is with longer TE.
second 180-degree pulse of a long TR spin echo pulse sequence, where the first echo
is proton density weighted, with short TE. T2 contrast differences are manifested by
allowing Mxy signal decay as shown in Figure 12-28. Compared with a T1-weighted
image, inversion of tissue contrast occurs in the image where CSF is bright, and gray
and white matter are reversed in intensity.
A T2-weighted image (Fig. 12-29) demonstrates high tissue contrast, compared
with either the T1-weighted or proton density-weighted images. As TE is increased,
more T2-weighted contrast is achieved, but at the expense of less Mxy signal and
greater image noise. However, even with low signal amplitudes, image processing
with window width and window level adjustments can remap the signals over the full
range of the display, so that the overall average brightness is similar for all images.
The typical T2-weighted sequence uses a TR of approximately 2,000 to 4,000 ms and
a TE of 80 to 120 ms.
FIGURE 12-29 T2 contrast weighting. Long TR minimizes

T1 relaxation differences of the tissues. Long TE allows T2
decay differences to be manifested. A second echo provides
time for T2 decay to occur, so a T2 W image is typically
acquired in concert with a PD W image. While this sequence
has high contrast, the signal decay reduces the overall signal
and therefore the SNR.
427
TABLE 12-5 SE PULSE SEQUENCE CONTRAST WEIGHTING PARAMETERS

PARAMETER
TR (ms)
TE (ms)
T1 CONTRAST
PROTON DENSITY CONTRASTa
T2 CONTRAST
400600
2,0004,000
2,0004,000
530
60150
530
Strictly speaking, SE images with TR less than 3,000 ms are not proton density with respect to the CSF;
because of its long T1, only 70% of the CSF magnetization recovery will have occurred and will not appear
as bright as for a true PD image. True PD image intensities can be obtained with fast spin echo methods
(Chapter 13) with longer TR (e.g., 8,000 ms).
TE, time of echo; TR, time of repetition.
Spin Echo Parameters

Table 12-5 lists typical contrast-weighting values of TR and TE for SE imaging. For
conventional SE sequences, both proton density and T2-weighted contrast signals
are acquired during each TR by acquiring two echoes with a short TE and a long TE
(Fig. 12-30).
Inversion Recovery
IR emphasizes T1 relaxation times of the tissues by extending the amplitude of the longitudinal recovery by a factor of two. An initial 180-degree RF pulse inverts Mz to 2Mz.
After a programmed delay, the time of inversionTI, a 90-degree RF (readout) pulse
rotates the recovered fraction of Mz into the transverse plane to generate the FID.
A second 180-degree pulse (or gradient polarity reversal, see next section, GE) at
TE/2 produces an echo at TE (Fig. 12-31); in this situation, the sequence is called
inversion recovery spin echo (IR SE). The TR for IR is the time between 180-degree
initiation pulses. Partial saturation of the protons and steady-state equilibrium of
the longitudinal magnetization is achieved after the first three excitations in the
sequence. The echo amplitude associated with a given tissue depends on TI, TE, TR,
and magnitude (positive or negative) of Mz.
The signal intensity at a location (x,y) in the image matrix for an IR SE acquisition
with nonmoving anatomy is approximated as
S rH (1 2 2e2TI/T1) (1 2e2(TR2TI)/T1) (e2TE/T2)
0
10 ms
45 ms
90
180
180
TE1= 20 ms
1st echo
TR
2500 ms
90
TE2 = 90 ms
2nd echo
FIGURE 12-30 SE with two 180-degree refocusing pulses after the initial 90-degree excitation pulse.
The early echo contains information related to proton density of the tissues, and the longer echo provides
T2 weighting. This double echo method is used for providing proton density content and T2 weighted content
independently during the same TR interval, and used to fill two separate k-space repositories that are used in
producing the final proton density and T2 weighted images.
428
TR
TI
TE
180
90
180
180
excitation
readout
refocusing pulse
excitation
RF pulses
TE/2
Signal out
FIGURE 12-31 Inversion-recovery SE sequence is shown. The initial 180-degree excitation pulse inverts
the longitudinal magnetization, and thus requires a factor of two times recovery of the longitudinal
magnetization over time. The inversion time (TI) is the delay between the excitation pulse and conversion to transverse magnetization of the recovered longitudinal magnetization. Subsequently, a second
180-degree pulse is applied at TE/2, which refocuses the transverse magnetization as an echo at time
TE. The signal strength is chiefly a function of the T1 characteristics of the tissues, as the TE values are
keptshort.
where the factor of 2 in the first part of the equation arises from the longitudinal
magnetization recovery from 2Mz to Mz, and all other parameters are as previously
described. For the TI to control contrast between tissues, it follows that TR must be
relatively long and TE short. The RF sequence is shown in Figure 12-31.
The IR sequence produces negative longitudinal magnetization that results in
negative (in phase) or positive (out of phase) transverse magnetization when short
TI is used. The actual signals are acquired as magnitude (absolute values) such that
Mz values are positive.
Short Tau Inversion Recovery

Short Tau Inversion Recovery, or STIR, is a pulse sequence that uses a very short TI and
magnitude signal processing, where Mz signal amplitude is always positive (Fig. 12-32).
Inversion Recovery (T1)
Fat
Image
intensity
Fat
White
Gray
White
Gray
CSF
CSF
Time (ms))
Time (ms)
TE
-M z
180
pulse
TI
90
pulse
Signal intensity is dependent on T1 (with short TE)

T1 relaxation range is doubled
Contrast is dependent on TI
RF energy depos ition is relatively high
FIGURE 12-32 The IR longitudinal recovery diagram shows the 2 3 Mz range provided by the 180-degree
excitation pulse. A 90-degree readout pulse at a time TI and a 180-degree refocusing pulse at a time TE/2 from
the readout pulse forms the echo at time TE. The time scale is not explicitly indicated on the x-axis. A short TE
is used to reduce T2 contrast characteristics.

Magnitude of Mz (T1)
Mz
429
Image
intensity
Mxy
CSF
CSF
Gray
Gray
Fat
Fat
Time (ms))
Time (ms))
bounce point
TI
STIR
TE
FIGURE 12-33 IR longitudinal magnetization as a function of time, with magnitude signal processing. All
tissues go through a null (the bounce point) at a time dependent on T1. The inversion time (TI) is adjusted to
select a time to null a certain tissue type. Shown above is the STIR (Short Tau IR) used for suppressing the signal
due to fat tissues, achieved with TI = approximately 150 ms (0.693 3 260 ms).
In this situation, materials with short T1 have a lower signal intensity (the reverse of
a standard T1-weighted image), and all tissues at some point during recovery have
Mz 5 0. This is known as the bounce point or tissue null. Selection of an appropriate
TI can thus suppress tissue signals (e.g., fats/lipids, CSF) depending on their
T1 relaxation times. The signal null (Mz 5 0) occurs at TI 5 ln(2) 3 T1, where ln is
the natural log and ln(2) 5 0.693. Since T1 for fat at 1.5T is approximately 260 ms,
TI is selected as 0.693 3 260 ms 5 180 ms. A typical STIR sequence uses TI of 140
to 180 ms and TR of approximately 2,500 ms. Compared with a T1-weighted examination, STIR reduces distracting fat signals (Fig. 12-33) and chemical shift artifacts
(explained in Chapter 13) (Fig. 12-34).
Fluid Attenuated Inversion Recovery

The signal levels of CSF and other tissues with long T1 relaxation constants can be
overwhelming in the magnitude IR image. Fluid attenuated IR, the FLAIR sequence,
T1
STIR
FIGURE 12-34 SE T1-weighting versus STIR technique. Left. T1 W with TR = 750 ms, TE = 13 ms. Right.
STIR with TR = 5,520 ms, TI = 150 ms, TE = 8 ms. The fat is uniformly suppressed in the STIR image, providing
details of nonfat structures otherwise difficult to discern.
430
Magnitude of Mz (T1)
Mz

M xy
Image
intensity
Fat
Gray
Fat
Gray
CSF
CSF
Time (ms))
Time (ms))
bounce point
TI
TE
FLAIR
FIGURE 12-35 Shown above is the FLAIR acquisition, with the TI set to the null of CSF, which reduces the
large CSF signals and allows the visualization of subtle details otherwise hidden. TE is short to not allow T2
contamination of the signals.
reduces CSF signal and other water-bound anatomy in the MR image by using a
TI selected at or near the bounce point of CSF to permit better evaluation of
the surrounding anatomy as shown in Figure 12-35. Reducing the CSF signal
(T1 < 2,500 ms @ 1.5 T) requires TI 5 0.693 3 2,500 ms, or < 1,700 ms. A
comparison of T1, T2, and FLAIR sequences demonstrates the contrast differences achievable by reducing signals of one tissue to be able to visualize another
(see Fig. 12-36).
Summary, Spin Echo Sequences

MR contrast schemes for clinical imaging use the SE or an IR variant of the SE pulse
sequences for many examinations. SE and IR SE sequences are less sensitive to magnetic field inhomogeneities, magnetic susceptibilities, and generally give high SNR
and CNR. The downsides are the relatively long TR and corresponding long acquisition times.
Gradient Echo
The GE technique uses a magnetic field gradient applied in one direction and
then reversed to induce the formation of an echo, instead of the 180-degree
T1
T2
FLAIR
FIGURE 12-36 Left. T1-weighted spin-echo axial brain image (TR = 549 ms, TE = 11 ms); Middle: T2 weighted
spin-echo image (TR = 2,400 ms, TE = 90 ms); Right. FLAIR image (TR = 10,000 ms, TI = 2,400 ms, TE = 150 ms).
431
inversion pulse. For a FID signal generated under a linear gradient, the transverse
magnetization dephases rapidly as the gradient is applied. After a predetermined
time, the nearly instantaneous reversal of the GE polarity will rephase the protons and produce a GE that occurs when the opposite gradient polarity of equal
strength has been applied for the same time as the initial gradient. The induced
GE signal is acquired just before and after the peak amplitude, as illustrated in
Figure12-37.
The GE is not a true SE but a purposeful dephasing and rephasing of the FID.
Magnetic field (B0) inhomogeneities and tissue susceptibilities caused by paramagnetic or diamagnetic tissues or contrast agents are emphasized in GE imaging.
This is because the dephasing and rephasing of the FID signals occur in the same
direction relative to the main magnetic field. In this situation, unlike a 180-degree
refocusing RF pulse, the external magnetic field variations are not cancelled. Compare Figures 12-22 and 12-37 and the direction of the Mxy vector in the rotating
frame. For SE techniques, the FID and echo vectors form in opposite directions,
whereas with GE, the FID and echo form in the same direction. Significant sensitivity to field nonuniformity and magnetic susceptibility agents thus occurs, as
Mxy decay is a strong function of T2*, which is much shorter than the true T2
achieved in SE sequences. Timing of the gradient echo is controlled either by
inserting a time delay between the negative and positive gradients or by reducing
the amplitude of the reversed gradient, thereby extending the time for the rephasing process to occur.
A major variable determining tissue contrast in GE sequences is the flip angle.
Depending on the desired image contrast, flip angles of a few degrees to more than
90 degrees are used, a majority of which are small angles much less than 60 degrees.
In the realm of very short TR, smaller flip angles require less time and create more
transverse magnetization compared to larger flip angles. A plot of Mxy signal amplitude versus TR as a function of flip angle (Fig. 12-38) shows that smaller flip angles
produce more Mxy signal than larger flip angles when TR is less than 200 ms.
Excitation
<60
Gradient
Gradient Echo
Equal area, opposite polarity
+
TE
Mxy
TE/2
Rotating frame
Mxy reforms in
same direction
Gradient dephasing
Gradient rephasing
FIGURE 12-37 A magnetic field gradient induces the formation of an echo (instead of a 180-degree RF
pulse). Transverse magnetization spins are dephased with an applied gradient of one polarity and rephased
with the gradient reversed in polarity; this produces a gradient echo. Note that the rotating frame depicts
the magnetic moment vector of the echo in the same direction as the FID relative to the main magnetic field,
and therefore extrinsic inhomogeneities are not cancelled.
432
Normalized signal amplitude:

Transverse magnetization
1.0
Flip angle
Flip angle
90
0.8
30 50 90
0.2
20
50
0.6
10
0.1
30
0.4
20
0.2
10
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
TR (sec)
0.0
0.25
TR (sec)
FIGURE 12-38 Transverse magnetization as a function of TR and the flip angle is shown. For small flip
angles and very short TR, the transverse magnetization is higher for small flip angles compared to larger flip
angles. Detail is shown on the right.
ltimately, tissue contrast in GE pulse sequences depends on TR, TE, and flip angle
U
along with specialized manipulation of the acquisition sequence.
Gradient Echo Sequences with Long TR

For GE sequences with long TR (greater than 200 ms) and flip angles greater than
45 degrees, contrast behavior is similar to that of SE sequences. The major difference
is image contrast that is based on T2* rather than T2, because external magnetic field
inhomogeneities are not cancelled. As for clinical interpretation, the mechanisms
of contrast based on T2* are different from those based on T2, particularly for MRI
contrast agents. A relatively long TE tends to emphasize the differences between T2*
and T2, rather than improve T2 contrast, as would be expected in an SE sequence.
T1 weighting is achieved with a short TE (5 to 10 ms). In most situations, however,
GE imaging is not useful with long TR, except when contrast produced by magnetic
susceptibility differences is desired.
Gradient Echo Sequences with short TR, less than 50 ms

Reducing the repetition period below 50 ms does not allow for transverse decay
(T2*) to fully occur, and a steady-state equilibrium of longitudinal and transverse
magnetization from pulse to pulse exists. The persistent transverse magnetization is
produced from previous RF excitations, and multiple signals are generated: (1) the
FID signal generated at the end of the current RF pulse, and once rephased, contains
T2* or T1 information depending on the TE; (2) a stimulated echo generated from
the previous RF pulse acting on the persistent transverse magnetization accruing
from the RF pulse twice displaced. The stimulated echo contains T2 and T2* weighting. This is schematically shown in Figure 12-39 for a train of RF pulses and the
generated signals.
Given the nature of the overlapping signals, there are generic GE acquisitions
termed coherent, incoherent, and steady-state free precession (SSFP) that provide
differential tissue contrast weighting.
Coherent GE
The coherent GE sequence is shown in Figure 12-40, indicating the timing of the
RF pulse with the dephasing and rephasing implemented by reversal of gradient
RF 1
TR
RF 2
FID 1
RF 3
FID 2
433
RF 4
FID 3
FID 4
SE 2
SE 1
FIGURE 12-39 GE acquisition with short TR less than 50 ms and flip angles upto 45 degrees produces two
signals: (1) the FID from the current RF pulse and (2) stimulated echo from the previous RF pulse resulting from
persistent transverse magnetization. These signals overlap, but are shown distinct in the illustration for clarity.
Resultant image contrast is due to the ratio of T1 to T2 in a tissue because of the mixed signals generated by
the combined FID (chiefly T1 and T2* contrast) and SE (chiefly T2 and T2* contrast) in the digitized signal. With
the train of RF excitation pulses, RF pulse 2 stimulates the echo formation of the FID produced from RF pulse 1,
which appears during RF pulse 3 and superimposes on the current (RF pulse 3) FID. While this is a conceptual
illustration, in fact, the actual situation is much more complicated, as there are many higher order stimulated
and gradient echoes that contribute to the observed signal.
polarity to generate an echo at a selectable time TE for the frequency encode gradient (FEG), where identification of proton position based upon frequency is performed. (Note: the various encoding gradients used to localize the protons are
discussed in the next section of this chapter.) Also involved in this sequence is the
phase encode gradient (PEG), which is applied and incrementally changed for each
TR to identify proton position in the direction perpendicular to the FEG based
upon phase changes of the protons after the PEG is turned off. A rewinder PEG
of the same strength but opposite polarity is applied after the echo to realign the
phase prior to the next RF pulse. The combined FID and simulated echo signals
are generated during the GE, and produce tissue contrast dependent on flip angle,
TR, and TE. With moderate flip angles of 30 to 60 degrees, the differences in tissue contrast are primarily based upon T2/T1 ratios. Since most tissues with a long
TR
RF
rewinder
PEG
FEG
rephase
dephase
FID
SE
TE
Gradient Echo + Stimulated Echo
FIGURE 12-40 Coherent GE: Multiple gradients are used for localizing the protons, including the PEG and
the FEG. In addition to localizing protons, these gradients are intrinsic to the manipulation of the signals in
the coherent GE acquisition. As shown above, the PEG is applied after the RF pulse but prior to the FEG, to
determine location in one direction by adding a known phase to the protons based on location per TR interval.
Subsequently, a rewinder gradient of the same strength but opposite polarity resets the phase prior to the
next TR excitation. The FEG is the gradient that generates the echo from the FID plus the stimulated echo from
the previous RF pulse.
434
TABLE 12-6 GRADIENT RECALLED ECHO WEIGHTING (STEADY-STATE)

PARAMETER
Flip angle
(degrees)
TR (ms)
TE (ms)
T1
T2/T1
T2
T2*
4590
3050
515
515
200400
1050
200400
100300
100300
315
3050
1020
515
315
PROTON DENSITY
530
T2 also have a long T1 and vice versa, there is very little tissue contrast generated.
In certain applications such as MR angiography, this is a good outcome, so that anatomical contrast differences that would otherwise compete with the bright blood
when displayed by image processing techniques (described in Chapter 13) would
reduce the image quality.
This acquisition technique is described by the acronyms GRASS (gradient
recalled acquisition in the steady state), FISP (fast imaging with steady-state
precession), FAST (Fourier acquired steady state), and other acronyms coined by
MRI equipment manufacturers. Table 12-6 indicates typical parameter values for
contrast desired in steady-state and GE acquisitions. The steady-state sequences
with coherent echoes produce T2* and proton density contrast. A GRASS/FISP
sequence using TR 5 35 ms, TE 5 3 ms, and flip angle 5 20 degrees shows unremarkable contrast, but blood flow shows up as a bright signal (Fig. 12-41). This
technique enhances vascular contrast, from which MR angiography sequences can
be reconstructed.
Incoherent, Spoiled Gradient Echo Techniques

With very short TR steady-state acquisitions, T1 weighting cannot be achieved to
any great extent, owing to either a small difference in longitudinal magnetization
with small flip angles or dominance of the T2* effects for larger flip angles produced
TR=35 ms, TE=3 ms, flip angle=20
FIGURE 12-41 A steady-state gradient recalled echo (GRASS) sequence (TR = 24 ms, TE = 4.7 ms, flip
angle = 50 degrees) of two slices out of a volume acquisition is shown. Contrast is unremarkable for white
and gray matter because of a T2-/T1-weighting dependence. Blood flow appears as a relatively bright signal.
MR angiography (see Chapter 13) depends on pulse sequences such as these to reduce the contrast of the
anatomy relative to the vasculature.

RF 1, phase 0
RF 2, phase 45
FID 1, phase 0
RF 3, phase 90
FID 2, phase 45
435
RF 4, phase 135
FID 3, phase 90
FID 4, phase 135
SE 1, Phase 0
SE 2, phase 45
FIGURE 12-42 Incoherent (spoiled) GE acquisition: Persistent transverse magnetization is dephased by each
RF pulse; the superposition of the FID from the current RF pulse with a stimulated-echo from the previous RF
pulse is separable, because of differences in the phase of the generated signals. In an actual sequence with TR
of 5 to 6 ms, several of stimulated echoes will contribute to the signal, along with the FID. Particular angles
of phase increment (typically 117 or 123 degrees) have been determined empirically to cause cancellation by
destructive interference of the magnetization from the different coherence pathways to eliminate the signal,
leaving only the signal from the FID. T1 contrast can be preferentially generated without contamination of the
signals due to the T2 characteristics of the tissues.
by persistent residual transverse magnetization created by stimulated echoes. The

T2* influence can be reduced by using a long TR (usually not an option), or by
spoiling the steady-state transverse magnetization by introducing incoherent
phase differences from pulse to pulse. The latter is achieved by adding a phase
shift to successive RF pulses during the excitation of protons. Both the RF transmitter and RF receiver are phase locked, so that the receiver discriminates the phase
of the GE from the SE generated by the previous RF pulse, now out of phase, as
shown in Figure 12-42. Mostly, T1-weighted contrast is obtained, with short TR,
short TE, and moderate to large flip angle. Spoiled transverse magnetization gradient recalled echo (SPGR) is often used in three-dimensional volume acquisitions
because of the extremely short acquisition time allowed by the short TR of the GE
sequence and the good contrast rendition of the anatomy provided by T1 weighting (Fig. 12-43).
TR=8 ms, TE=1.9 ms, flip angle=20

Note T1 contrast and high blood vessel signal
FIGURE 12-43 Incoherent (spoiled) GE images. The ability to achieve T1 contrast weighting is extremely
useful for rapid three-dimensional volume imaging. Bright blood (lower portion of each image) and magnetic
susceptibility artifacts are characteristic of this sequence. TR = 8 ms, TE = 1.9 ms, flip angle = 20 degrees.
436
MR contrast agents (e.g., gadolinium) produce greater contrast with T1-weighted

SPGR than with a comparable T1-weighted SE sequence because of the greater sensitivity to magnetic susceptibility. The downsides of spoiled GE techniques are the
increased sensitivity to other artifacts such as chemical shift and magnetic field inhomogeneities, as well as lower SNR.
Steady-State Free Precession

In GE sequences with short TR (less than 50 ms), the TE is not long enough to generate any T2 contrast, and GE rephasing is inefficient and dominated by T2* effects.
The SSFP sequence emphasizes acquisition of only the stimulated echo, which arises
from the previous RF pulse and appears during the next RF pulse at a time equal to
2 3 TR (see Fig. 12-39). To reposition the stimulated echo to an earlier time in the
TR interval for signal acquisition, a rewinder gradient is applied, which speeds up the
rephasing process initiated by the RF pulse, so that it reforms and decays just before
the next RF pulse. In the SSFP sequence, there are two TE values: one is the actual
TE, the time between the peak stimulated echo amplitude and the next excitation
pulse, and the other is the effective TE, the time from the echo and the RF pulse that
created its FID. The effective TE is thus longer than the TR and is calculated as: effective TE 5 2 3 TR-actual TE. So for TR 5 50 ms and TE 5 5 ms, the effective TE is
95 ms, which means that the stimulated echo has had 95 ms to manifest T2 weighting. SSFP sequences provide true T2 contrast weighting, are useful for brain and
joint imaging, and can be used with three-dimensional volumetric acquisitions (as
discussed in Chapter 13) (Figs. 12-44 and 12-45). Balanced SSFP sequences generate accumulated gradient echo (FID) and stimulated echo signals with the use of
symmetrical gradients in 3 spatial directions, which null phase shifts induced by
flow. Balanced SSFP provides T2/T1 contrast and high speed acquisitions, particularly useful for cardiac imaging (Chavan, et.al., 2008).
Summary, Gradient Echo contrast

For GE acquisition in the realm of short TR, persistent transverse magnetization
produces two signals: (1) the FID produced from the RF pulse just applied and
RF 1
RF 2
FID 1
RF 3
RF 4
Rewind 1
SE 1
Rewind 2
SE 2
FID 2
ID 3
Rewind 3
SE 3
FID 4
Effective TE
Actual
TE
FIGURE 12-44 SSFP uses a rewinder gradient to position each stimulated echo so that it appears just before
the next excitation pulse, and is acquired separately from the FID. Since rephasing of the stimulated echo is
initiated by an RF pulse rather than a gradient, more T2 rather than T2* weighting occurs. The actual TE is the
time from the peak echo amplitude to the center of the next RF pulse. The effective TE is 2 TR2actual TE
(This is shown in the lower left of the figure).
437
FIGURE 12-45 Three abdominal post gadolinium contrast axial images are part of a breath-hold dataset
using a balanced SSFP acquisition, which simultaneously accumulates the FID and the stimulated echo
with contrast varying according to the T2/T1 ratio (TR: 3.34 ms, TE: 1.2 milliseconds, Flip angle 70 degrees,
matrix size 192 320). Each image is acquired in approximately 700 milliseconds, making this sequence
useful for reducing voluntary and involuntary patient motion in contrast enhanced abdominal and cardiac
imaging.
(2) the stimulated echo from the residual transverse magnetization. From these two
generated signals, coherent, incoherent, or SSFP sequences are used in order to generate different contrast-weighting schemes. Coherent GE uses the combined FID and
SE signals and produces contrast mainly depending on the ratio of T2/T1 or T2*/T1;
incoherent (or spoiled) GE produces contrast mainly based on T1, and SSFP samples
the SE signal only, thus producing signals that are more T2 weighted. Figure 12-46
shows the signals and the timing of acquisition for these three scenarios. Note that
these are generic, not fully inclusive, and do not include specific sequences implemented by the major manufacturers that do not fit into this simplified categorization.
Additionally the term SSFP is used differently by the manufacturers. One should
consult the specific applications manuals and descriptions of the specific models and
pulse sequence implementations for more detailed information (Also see Chapter
13, Table 1 for a brief comparison of GE sequences). Details of GE techniques are
reviewed in an article by Chavhan, et.al. (2008).
RF
pulse
FID + SE
FID
Coherent
GE
SE
TE
RF
pulse
Incoherent,
spoiled
FID
GE
TE
RF
pulse
FID
SSFP
SE
TE
FIGURE 12-46 A schematic summary of the three major GE

sequences (RF, RadioFrequency; FID, Free Induction Decay; GE,
Gradient Echo; SE, Stimulated or Spin Echo; TE, Time of Echo) are
shown. Coherent GE uses signals from the FID and the SE to produce the image, typically with contrast dependent on T2/T1 weighting, and low contrast. Incoherent GE eliminates the detection of
the SE, thus, providing a means to generate T1-weighted contrast
from the FID signal. SSFP uses the SE signal, which provides mainly
T2-weighted contrast. Balanced SSFP uses both the gradient and
stimulated echo to produce a T2/T1 weighting with symmetrically
applied gradients in three dimensions.
438
FIGURE 12-47 The gradient magnetic field

creates a net positive and negative magnetic
environment that adds to and subtracts from
the main magnetic field. Associated with the
local change in magnetic field is a positionally
dependent change in precessional frequencies,
per the Larmor equation. The precessional frequencies directly vary across the field in proportion to the applied gradient strength.
Lower
magnetic
field
Null
Higher
magnetic
field
+
Lower
precessional
frequency
Higher
precessional
frequency
12.6 MR Signal Localization

Spatial localization is essential for creating MR images and determining the
location of discrete sample volumes for MR spectroscopy. This is achieved by
superimposing linear magnetic field variations on the main (B0) field to generate corresponding position-dependent variations in precessional frequency of the protons
(Fig. 12-47). Simultaneous application of an RF excitation (B1 pulse) excites only
those protons in resonance within the frequency bandwidth (BW) of the B1 RF
pulse by absorbing energy.
Magnetic Field Gradients

Inside the magnet bore, three sets of gradients reside along the logical coordinate
axesx, y, and zand produce a magnetic field variation determined by the magnitude of the applied current in each coil set (Fig. 12-48). When independently
energized, the three coils (x, y, z) can produce a linearly variable magnetic field in
Individual gradients
Superimposed gradients
Y
X
X -axis
Z
B0
Y-axis
Net gradient =
G2x + G2y + G2z
Z -axis
FIGURE 12-48 Within the large stationary magnetic field, field gradients are produced by three separate
coil pairs placed within the central core of the magnet, along the x, y, or z directions. In modern systems, the
current loops are distributed across the cylinders for the x-, y- and z- gradients, which generates a lower, but
more uniform gradient field. Magnetic field gradients of arbitrary direction are produced by the vector addition
of the individual gradients turned on simultaneously. Any gradient direction is possible by superimposition of
magnetic fields generated by the three-axis gradient system.
439
any direction, where the net gradient is equal to G2x G2y G2z . Gradient polarity reversals (positive to negative and negative to positive changes in magnetic field
strength) are achieved by reversing the current direction in the gradient coils.
Two important properties of magnetic gradients are: (1) The gradient field strength
is determined by its peak amplitude and slope (change over distance), and typically
ranges from 1 to 50 mT/m. (2) The slew rate is the time to achieve the peak magnetic
field amplitude. Typical slew rates of gradient fields are from 5 to 250 mT/m/ms. As
the gradient field is turned on, eddy currents are induced in nearby conductors such
as adjacent RF coils and the patient, which produce magnetic fields that oppose the
gradient field and limit the achievable slew rate. Actively shielded gradient coils and
compensation circuits can reduce problems caused by eddy currents.
In a gradient magnetic field, protons maintain precessional frequencies corresponding to local magnetic field strength. At the middle of the gradient, called the null, there is no change in the field strength or precessional
frequency. With a linear gradient, the magnetic field increases and decreases
linearly from the null, as does the precessional frequency (Fig. 12-47 and
Table 12-7). In the rotating frame of reference, incremental changes of frequency occur symmetrically about the null, and the positions of protons are
encoded by frequency and phase. The frequency BW is the range of frequencies over the FOV, and the frequency BW per pixel is the BW divided by the
number of discrete samples. Gradient amplitude can also be expressed in frequency per distance. For instance, a 10-mT/m gradient can be expressed as
10 mT/m 3 42.58 MHz/T 3 1 T/1,000 mT 5 0.4258 MHz/m, which is equivalent to 425.8 kHz/m or 425.8 Hz/mm. The relationship of gradient strength
and frequency BW across the FOV is independent of the main magnet field
strength.
Localization of protons in the three-dimensional volume requires the application
of three distinct gradients during the pulse sequence: slice select, frequency encode,
and phase encode. These gradients are sequenced in a specific order, depending on
the pulse sequences employed. Often, the three gradients overlap partially or completely during the scan to achieve a desired spin state, or to leave protons in their
original phase state after the application of the gradient(s).
TABLE 12-7PRECESSIONAL FREQUENCY VARIATION AT 1.5 T ALONG
AN APPLIED GRADIENT
Gradient field strength
3 mT/m 5 127.74 Hz/mm
Main magnetic field strength
1.5 T
FOV
0.15 m 5 150 mm
Linear gradient amplitude over FOV
0.45 mT; from 20.225 mT to 10.225 mT
Maximum magnetic field (frequency)
1.500225 T (63.8795805 MHz)
Unchanged magnetic field at null
1.500000 T (63.8700000 MHz)
Minimum magnetic field
1.499775 T (63.8604195 MHz)
Net frequency range across FOV
0.019161 MHz 5 19.2 kHz 5 19,161 Hz
Frequency range across FOV (1,278 Hz/cm)b
127.74 Hz/mm 3 150 mm 5 19,161 Hz
Frequency BW per pixel (256 samples)
19,161 Hz/256 5 74.85 Hz/pixel
Calculated using the absolute precessional frequency range: 63.879663.8604 MHz.

Calculated using the gradient strength expressed in Hz/mm.
440
Slice Select Gradient

RF transmitters cannot spatially direct the RF energy to a specific region in the body;
rather the RF pulse, when turned on during the application of the slice select gradient (SSG), determines the slice location of protons in the tissues that absorb energy.
For axial MR images, the SSG is applied along the long (cranial-caudal) axis of the
body. Under the influence of the gradient field, the proton precessional frequencies
in the volume are incrementally increased or decreased dependent on their distance
from the null. A selective, narrow band RF frequency pulse of a known duration
and amplitude delivers energy to the total volume, but only those protons with precessional frequencies matching the RF BW frequencies will absorb energy within a
defined slice of tissues as shown in Figure 12-49 (red vertical line indicates energy
absorbed due to resonance). If the center frequency of the RF pulse is increased, then
a different slab of protons absorbs energy (blue vertical line).
Slice thickness is chiefly determined by the frequency BW of the RF pulse and
the gradient strength across the FOV. For a fixed gradient strength, the RF pulse with
a narrow BW excites protons within a thin slice, and a wide BW excites a thick slice
(Fig. 12-50A). For a fixed RF BW, a high gradient strength produces a large range of
frequencies across the FOV and decreases the slice thickness, whereas a low gradient
strength produces a small range of frequencies and produces an increase in the slice
thickness (Fig. 12-50B).
A combination of a narrow BW and a low gradient strength or a wide BW and a high
gradient strength can result in the same slice thickness. Decisions depend chiefly on the
SNR and the propensity for chemical shift artifacts. SNR is inversely proportional to
1
the receiver BW: SNR
; therefore narrow BW and low gradient strength are
BW
preferred; however, artifacts due to chemical shift (see Chapter 13, Section 13.5, Chemical Shift Artifacts) can be problematic. Consequently, trade-offs in image quality must be
considered when determining the optimal RF BW and SSG field strength combinations.
After the SSG is turned off, the protons revert to the precessional frequency of
the main magnetic field, but phase coherence is lost. To reestablish the original phase
of all stationary protons, a gradient of opposite polarity equal to one-half of the
Narrow -band RF:
Lower frequenc y
Higher frequency
Slice -select gradient
B0
y
z
Gradient coils
FIGURE 12-49 The SSG disperses the precessional frequencies of the protons in a known way along the gradient. A narrow band RF pulse excites only a selected volume (slice) of tissues, determined by frequency, BW, and SSG
strength. In the example above, two narrow-band RF pulses with different center frequencies irradiate the whole
body during the application of the gradient, and only those protons at the same frequencies as the RF pulses will
absorb energy. Note that the higher frequency slice is shifted towards the positive pole of the applied gradient.
Variable RF Bandwidth
Fixed Gradient
Fixed RF Bandwi dth,

Variable Gradient
Bandwidth
RF Frequency
441
Gradient Strength
fixed
Narrow Wide
gradient
fixed
Low
0
Large
Large
Small
Small
Slice Thickness, z
Narrow bandwidth
Small z
Wide bandwidth
High
Slice Thickness, z
High gradient
Small z
Large z
Low gradient
Large z
FIGURE 12-50 Slice thickness is dependent on RF BW and gradient strength. A. For a fixed gradient
strength, the RF BW determines the slice thickness. B. For a fixed RF BW, gradient strength determines the
slice thickness.
area of the original SSG is applied (Fig. 12-51). For 180-degree RF excitations, the
rephasing gradient is not necessary, as all protons maintain their phase relationships
because of the symmetry of the RF pulse and spin inversion.
To summarize, the SSG is applied simultaneously with a RF pulse of a known
BW to create proton excitation in a single plane with a known slice thickness, and
to localize signals orthogonal to the gradient. It is the first of three gradients applied
to the volume.
Frequency Encode Gradient

The FEG, also known as the readout gradient, is applied in a direction perpendicular to the SSG, along the logical x-axis, during the evolution and decay of the
induced echo. Net changes in precessional frequencies are distributed symmetrically
from 0 at the null to 1fmax and 2fmax at the edges of the FOV (Fig. 12-52) under
the applied FEG. The composite signal is amplified, digitized, and processed by the
Fourier transform to convert frequency into spatial position (Fig. 12-53). A spatial
projection is created by integrating the resultant Fourier transformed signal amplitudes perpendicular to the direction of the applied gradient at corresponding spatial
positions.
The SSG in concert with an incremental rotation of the FEG direction about
the object can produce data projections through the object as a function of angle,
as shown in Figure 12-54. With a sufficient number of projections, filtered backprojection can be used for reconstructing a tomographic image (see Chapter 11 on
RF pulse
Rephasing
Gradient
SSG
Equal area
Relative
phase
in phase
out of phase
in phase
FIGURE 12-51 SSG rephasing. At the peak of the RF

pulse, all protons in the slice are in phase, but the SSG
causes the spins to become dephased after the gradient
is turned off. To reestablish phase coherence, a reverse
polarity gradient is applied equal to one-half the area of
the original SSG. At the next pulse, the relative phase
will be identical.
442

Body cross -section
determined by SSG
Frequency Encode Gradient (FEG)

applied during echo formation
y
+
Net frequency at
gradient location
Composite
frequency
signal
signal
amplitude
Receiver coil
Digitization
- f max
+ f max
Fourier Transform
(position decoder)
Spatial position
Line integral of signal amplitude
FIGURE 12-52 The FEG is applied in an orthogonal direction to the SSG, and confers a spatially dependent
variation in the precessional frequencies of the protons. Acting only on those protons in a slice determined
by the SSG excitation, the composite signal is acquired, digitized, demodulated (Larmor frequency removed),
and Fourier transformed into frequency and amplitude information. A one-dimensional array represents a
projection of the slice of tissue (amplitude and position) at a specific angle. (Demodulation into net frequencies
occurs after detection by the receiver coil; this is shown in the figure for clarity only.)
CT Reconstruction). In fact, this is how some of the first MR images were reconstructed from individual projections, using a rotating FEG. However, inefficient
acquisition and data handling, in addition to motion artifacts, has led to near-universal acquisition with phase encoding techniques.
Demodulated MR Signal
Corresponding Fourier Transformation
2
1.5
1 cycle/cm
Amplitude: 1.0
1
0.5
0
-0.5
-1
1.0
0.5
-1.5
-2
-5 -4 -3 -2 -1
2
1.5
1
2 cycles/cm
cycles/
Amplitude: 0.75
180 phase shift
180
0.5
0
Amplitude
-0.5
-1
-1.5
-2
1.0
0.5
-5 -4 -3 -2 -1
2
1.5
1
0.5
3 cycles/cm
cycles/
Amplitude: 0.5
0
-0.5
-1
1.0
0.5
-1.5
-2
-5 -4 -3 -2 -1
2
1.5
1
Composite
waveform
0.5
0
-0.5
-1
1.0
0.5
-1.5
-2
1.0
Frequency (cm-1)
-5 -4 -3 -2 -1
Position, cm
FIGURE 12-53 Spatial frequency signals (cycles/cm) and their Fourier transforms (spatial position) are shown
for three simple sinusoidal waveforms with a specific amplitude and phase. The Fourier transform decodes
the frequency, phase and amplitude variations in the spatial frequency domain into a corresponding position
and amplitude in the spatial domain. A 180-degree phase shift (second from the top) is shifted in the negative
direction from the origin. The composite waveform (a summation of all waveforms, lower left) is decoded by
Fourier transformation into the corresponding positions and amplitudes (lower right).
FIGURE 12-54 A rotating FEG in the x-y plane

allows the acquisition of individual projections as
a function of angle, by repeating the SSG FEG
sequence with incremental change of the FEG
direction. This example shows eight projections;
to have sampling sufficient for high SNR and
resolution would require hundreds of projections
about the object.
2
4
4
....
443
3
2
Phase Encode Gradient

Position of the protons in the third orthogonal dimension is determined with a PEG,
which is applied after the SSG but before the FEG, along the third orthogonal axis. Phase
in this context represents a linear variation in the starting point of sinusoidal waves that
precess at the same frequency. Phase changes are purposefully introduced by the application of a short duration PEG within each data acquisition (DAQ) interval. Prior to the
PEG, all protons have the same phase, and turning on the PEG introduces a linear variation in the precessional frequency of the protons according to their position along the
gradient. After a brief interval, the PEG is turned off, all protons revert to the Larmor frequency, and linear phase shifts are manifested by a specific gradient strength and polarity.
Incremental positive phase change is introduced for protons under the positive pole,
negative phase change under the negative pole, and no phase change occurs for protons
at the null of the PEG. Throughout the acquisition sequence, the PEG strength and polarity is incrementally changed to introduce specific known phase changes as a function of
position across the FOV for each acquisition interval (Fig. 12-55). Spatial encoding is
determined by the amount of phase shift that has occurred. Protons located at the center
PEG
strength
+ #1
Phase shift in TR interval

after gradient removed
x
#60
#128
#188
#256
FIGURE 12-55 The PEG is applied before the FEG and after the SSG. The PEG produces a spatially dependent
variation in angular frequency of the excited spins for a brief duration, and generates a spatially dependent
variation in phase when the spins return to the Larmor frequency. Incremental changes in the PEG strength for
each TR interval spatially encodes the phase variations: protons at the null of the PEG do not experience any
phase change, while protons in the periphery experience a large phase change dependent on their distance
from the null. The incremental variation of the PEG strength can be thought of as providing specific views
of the volume because the SSG and FEG remain fixed throughout the acquisition.
444
TR
90
180
RF
90
Excite protons
SSG
Fat Localize (z)
PEG
Localize (y)
FEG
Localize (x)
Echo
Generate echo
DAQ
Acquire data
TE
FIGURE 12-56 A typical spin-echo pulse sequence diagram indicates the timing of the SSG, PEG, and FEG
during the repetition time (TR) interval, synchronized with the RF pulses and the DAQ when the echo appears.
Each TR interval is repeated with a different PEG strength (this appears as multiple lines in the illustration, but
only one PEG strength is applied per TR as indicated by the bold line in this figure).
of the FOV, the PEG null, do not exhibit a phase shift. Protons located at the edges of the
FOV exhibit the largest positive to negative (or negative to positive) phase shifts. Protons
located at intermediate distances from the null experience intermediate phase shifts (positive or negative). Each location along the phase encode axis is spatially encoded by the
amount of phase shifts experienced by the protons. For sequential acquisition sequences,
each sample in the PEG direction is separated in time by the TR interval.
Gradient Sequencing
An acquisition of an SE pulse sequence is illustrated in Figure 12-56, showing the
timing of the SSG in conjunction with the 90-degree RF excitation pulse, the application of a short duration PEG at a known strength, followed by a 180-degree refocusing RF pulse at TE/2, and the echo envelope with the peak amplitude occurring at
TE. This sequence is repeated with slight incremental changes in the PEG strength to
define the three dimensions in the image over the acquisition time.
12.7 K-Space Data Acquisition and Image Reconstruction

MR data are initially stored in the k-space matrix, the frequency domain repository (Fig. 12-57). K-space describes a two-dimensional matrix of positive and negative spatial frequency values, encoded as complex numbers (e.g., a 1 bi, i 5 1 ).
The matrix is divided into four quadrants, with the origin at the center representing
frequency 5 0. Frequency domain data are encoded in the kx direction by the FEG,
and in the ky direction by the PEG in most image sequences. The lowest spatial
frequency increment (the fundamental frequency) is the BW across each pixel (see
Table 12-7). The maximum useful frequency (the Nyquist frequency) is equal to
frequency range across the kx or ky directions, as the frequencies are encoded
from2fmax to 1 fmax. The periodic nature of the frequency domain has a built-in symmetry described by symmetric and antisymmetric functions (e.g., cosine and sine
waves). Real, imaginary, and magnitude describe specific phase and amplitude
characteristics of the composite MR frequency waveforms. Partial acquisitions are
FIGURE 12-57 The k-space matrix is the repository for spatial frequency signals acquired during
the evolution and decay of the echo. The kx axis
(along the rows) and the ky axis (along the columns) have units of cycles/unit distance. Each axis
is symmetric about the center of k-space, ranging from fmax to +fmax along the rows and the
columns. The matrix is filled one row at a time in
a conventional acquisition with the FEG induced
frequency variations mapped along the kx axis and
the PEG induced phase variations mapped along
the ky axis.
Phase variations
+f max,y
ky
445
-f max, y
Frequenc y variations
-f max, x
+f max, x
kx
possible (e.g., one-half of the k-space matrix plus one line) with complex conjugate
symmetry filling the remainder (See Chapter 13, Section 1, Data Synthesis).
Two-Dimensional Data Acquisition

MR data are acquired as a complex, composite frequency waveform. With methodical
variations of the PEG during each excitation, the k-space matrix is filled to produce
the desired variations across the frequency and phase encode directions as shown in
Figure 12-58.
128 128 acquisition matrix
Repeat with different PEG for each TR

90
180
RF pulses
Slice select gradient
Phase encode gradient
Frequency encode gradient
MR signal
Data acquisition
z
0
64
2-D Fourier
ky
0
Transform
-63
-63
0
64
kx
Frequency Domain k-space
127
0
127
x
Spatial Domain image space
FIGURE 12-58 MR data are acquired into k-space matrix, where each row in k-space represents spatially
dependent frequency variations under a fixed FEG strength, and each column represents spatially dependent
phase shift variations under an incrementally varied PEG strength. Data are placed in a specific row determined
by the PEG strength for each TR interval. The grayscale image is constructed from the two-dimensional Fourier
transformation of the k-space matrix by sequential application of one-dimensional transforms along each row,
and then along each column of the intermediate transformed data. The output image matrix is arranged with
the image coordinate pair, x = 0, y = 0 at the upper left of the image matrix.
446
A summary description of the two-dimensional spin-echo image acquisition

steps follows.
1. A narrow band RF excitation pulse simultaneously applied with the SSG causes
a specific slab of tissues with protons at the same frequency to absorb energy.
Transverse magnetization, Mxy, is produced with amplitude dependent on the saturation of the protons and the angle of excitation. A 90-degree flip angle produces
the largest Mxy.
2. A PEG is applied for a brief duration, which introduces a phase difference among
the protons along the phase encode direction to produce a specific view of the
data along the ky axis, corresponding to the strength of the PEG,
3. A refocusing 180-degree RF pulse is delivered at TE/2 to invert and reestablish the
phase coherence of the transverse magnetization at time TE.
4. During the evolution and decay of the echo signal, the FEG is applied orthogonal
to both the SSG and PEG directions, generating spatially dependent changes in
the precessional frequencies of the protons.
5. Data sampling and acquisition of the complex signal occurs simultaneous to the
FEG. A one-dimensional inverse Fourier transform converts the digital data into
discrete frequency values and corresponding amplitudes to determine position
along the kx (readout) direction.
6. Data are deposited in the k-space matrix at a row location specifically determined by the strength of the PEG. For each TR, an incremental variation of the
PEG strength sequentially fills each row. In some sequences, the phase encode
data are acquired in nonsequential order to fill portions of k-space more pertinent to the requirements of the exam (e.g., in the low-frequency, central
area of k-space). Once filled, the k-space matrix columns contain positionally
dependent variations in phase change along the ky (phase encode) direction.
7. After all rows are filled, an inverse Fourier transform decodes the frequency
domain variations in phase for each of the columns of k-space to produce the
spatial domain representation.
8. The final image is scaled and adjusted to represent the proton density, T1, T2,
and flow characteristics of the tissues using a grayscale range, where each pixel
represents a voxel.
The bulk of image information representing the lower spatial frequencies is contained in the center of k-space, whereas the higher spatial frequencies are contained
in the periphery, as shown in Figure 12-59, representing a grayscale rendition
of k-space for a sagittal slice of a brain MR image acquisition. The innermost
areas represent the bulk of the anatomy, while the outer areas of k-space contain
the detail and resolution components of the anatomy, as shown by reconstructed
images.
Two-Dimensional Multiplanar Acquisition

Direct axial, coronal, sagittal, or oblique planes can be obtained by energizing the
appropriate gradient coils during the image acquisition, as shown in Figure 12-60.
The SSG determines the orientation of the slices: axial uses z-axis coils; coronal uses
y-axis coils; and sagittal uses x-axis coils for selection of the slice orientation. Oblique
plane acquisition depends on a combination of the x-, y-, and z-axis coils energized
simultaneously. SSG, PEG, and FEG applications are perpendicular to each other,
and acquisition of data into the k-space matrix remains the same, with the FEG along
the kx axis and the PEG along the ky axis.
128
Full k -space
Low & high frequenc ies
Central area
Low frequenc ies
447
Peripheral area
High frequencies
-127
-127
128
255
255
Spatial domain images reconstructed by 2D Fouri er Transform

FIGURE 12-59 A. Image representations of k-space segmentation show a concentration of information
around the origin (the k-space images are logarithmically amplified for display of the lowest amplitude signals).
B. Inverse two-dimensional Fourier transformation converts the data into a visible image. C. Segmenting a
radius of 25 pixels out of 128 in the central area and zeroing out the periphery extracts a majority of the lowfrequency information. D. The corresponding image demonstrates the majority of the image content is in the
center of k-space. E. Zeroing out the central portion and leaving the peripheral areas isolates the higher spatial
frequency signals. F. The resulting image is chiefly comprised of high frequency detail and resolution. Ringing
that is visible in the image is due to the sharp masking transition from the image data to zero.
12.8 Summary
The basics of magnetic resonance are covered in this chapter, including the simplest
descriptions of magnetism, magnetic characteristics of the elements, and magnetization of tissue samples. Important is the description of the intrinsic decay constants
T1, T2, T2*, and proton density in terms of tissue-specific structures and variation
in the intrinsic and extrinsic local magnetic fields. Contrast between tissues is determined by pulse sequences including SE, IR, GE, and their associated parameters
Axial
Coronal
Sagittal
FIGURE 12-60 Direct acquisitions of axial, coronal, and sagittal tomographic images are possible by electronically energizing the magnetic field gradients in a different order without moving the patient. Oblique
planes can also be obtained. PEG (ky axis) and FEG (kx axis) are perpendicular to the SSG.
448
TR, TE, TI, and flip angle. In addition to generating contrast, the ability to spatially
localize the protons and create a two-dimensional image is as important, so that the
differences can be appreciated in a grayscale rendition of the anatomy. The concepts
of the frequency domain description of the signals and the k-space acquisition matrix
are integral to the discussion, as well as the Fourier transform and the conversion of
frequency to spatial domain representations of the data, necessary for visualization.
In the next chapter, more details are given for image acquisition time, various pulse
sequence designs, how image acquisition time is shortened, and characteristics of
the image in terms of SNR, CNR, and artifacts. Unique capabilities for noninvasive
biopsy of tissues, quality control, equipment, MR safety, and biological effects are
also discussed.
REFERENCES AND Suggested Readings

Bottomley PA, Foster TH, Argersinger RE, Pfeifer LM. A review of normal tissue hydrogen NMR relaxation
mechanisms from 1-100 MHz: dependence on tissue type, NMR frequency, temperature, species,
excision, and age. Med Phys 1984;11:425448.
Chavhan GB, Babyn PS, Jankharia BG, Cheng HM,Shroff MM. Steady-State MR Imaging Sequences:
Physics, Classification, and Clinical Applications. Radiographics 2008;28:11471160.
Fullerton GD, Potter JL, Dornbluth NC. NMR relaxation of protons in tissues and other macromolecular
water solutions. Magn Reson Imaging 1982; 1:209228.
NessAiver M. All you really need to know about MRI physics. Baltimore, MD: Simply Physics, 1997.
Pooley RA. AAPM/RSNA physics tutorial for residents: fundamental Physics of MR imaging. Radiographics
2005;25:10871099.
Westbrook C, Kaut-Roth C, Talbot J. MRI in practice. 3rd ed. Malden, MA: Blackwell Publishing, 2005.
Chapter
13
Magnetic Resonance
Imaging: Advanced Image
Acquisition Methods, Artifacts,
Spectroscopy, Quality Control,
Siting, Bioeffects, and Safety
The essence of magnetic resonance imaging (MRI) in medicine is the acquisition,
manipulation, display, and archive of datasets that have clinical relevance in the
context of making a diagnosis or performing research for new applications and
opportunities. There are many advantages and limitations of MRI and MR spectroscopy (MRS) as a solution to a clinical problem. Certainly, as described previously
(note that this chapter assumes a working knowledge of Chapter 12 content), the
great advantages of MR are the ability to generate images with outstanding tissue
contrast and good resolution, without resorting to ionizing radiation. Capabilities of
MR extend far beyond those basics, into fast acquisition sequences, perfusion and
diffusion imaging, MR angiography (MRA), tractography, spectroscopy, and a host of
other useful or potentially useful clinical applications. Major limitations of MR are
also noteworthy, including extended acquisition times, MR artifacts, patient claustrophobia, tissue heating, and acoustic noise to name a few. MR safety, often ignored, is
also of huge concern to the safety of the patient.
In this second of two MR chapters, advanced pulse sequences and fast image
acquisition methods, dedicated radiofrequency (RF) coils, methods for perfusion,
diffusion, and angiography imaging, image quality metrics, common artifacts, spectroscopy, MR equipment and siting, as well as MR safety issues are described and
discussed with respect to the underlying physics.
The concepts of image acquisition and timing issues for standard and advanced
pulse sequences into k-space is discussed first, with several methods that can be used
to reduce acquisition times and many of the trade-offs that must be considered.
13.1 Image Acquisition Time

A defining character of MRI is the tremendous range of acquisition time needed to
image a patient volume. Times ranging from as low as 50 ms to tens of minutes are
commonly required depending on the study, pulse sequence, number of images in
the dataset, and desired image quality. When MR was initially considered to be a
potential diagnostic imaging modality in the late 1970s, the prevailing conventional
wisdom gave no chance for widespread applicability because of the extremely long
times required to generate a single slice from a sequentially acquired dataset, which
449
450
required several minutes or more per slice. Breakthroughs in technology, equipment design, RF coils, the unique attributes of the k-space matrix, and methods of
acquiring data drastically shortened acquisition times (or effective acquisition times)
quickly, and propelled the rapid adoption of MRI in the mid-1980s. By the early
1990s, MRI established its clinical value that continues to expand today.
Acquisition Time, Two-Dimensional Fourier Transform Spin

Echo Imaging
The time to acquire an image is determined by the data needed to fill the fraction
of k-space that allows the image to be reconstructed by Fourier transform methods.
For a standard spin echo sequence, the relevant parameters are the TR, number of
phase encoding steps, and number of excitations (NEX) used for averaging identical
repeat cycles, as
Acquisition time TR # PEG Steps NEX
Even though there may be multiple echoes as illustrated in Figure 13-1, there is
also the same number of k-space repositories to capture the data in a specific, single
row of k-space defined by the strength of the PEG, as shown for the first echo
with proton density weighting and second echo with T2 weighting for this double
echo acquisition. Thus, effective imaging time can be reduced by producing two
(or more) images of the same slice within the TR interval. In addition, the matrix
size that defines k-space is often not square (e.g., 256 256, 128 128), but
90
180
180
90
RF
SSG
PEG
FEG
Echo
DAQ
ky
ky
kx
st
1 echo, proton density weighted
kx
nd
2 echo, T2 weighted
FIGURE 13-1 Standard spin echo pulse sequence is shown with two echoes per TR interval to encode
proton density contrast (short TE, first echo), and T2 contrast (long TE, second echo). In this acquisition, two
separate images are acquired independently by storing in a designated k-space matrix according to echo time.
A single PEG strength is momentarily applied to induce phase variations to encode the row to be filled in each
of the matrices (see the red PEG encoding for the last row in k-space, for instance). The full k-space matrix
requires the sequence to be repeated with incremental variations in the PEG strength until each k-space row is
fully populated. If averaging is desired, then an identical sequence (without incrementing the PEG) is repeated
and averaged in the same row.
Chapter 13 Magnetic Resonance Imaging
451
r ectangular (e.g., 256 192, 256 128) where the small matrix dimension is most
frequently along the phase encode direction to minimize the number of incremental
PEG strength applications during the acquisition. A 256 192 image matrix and
two averages (NEX) per phase encode step with a TR 600 ms (for T1 weighting)
requires imaging time of 0.6 s 192 2 230.4 s 3.84 min for a single slice!
For a proton density and T2-weighted double echo sequence with TR 2,500 ms
(Fig. 13-1), this increases to 16 min, although two images are created in that time.
Of course, a simple first-order method would be to eliminate the number of averages (NEX), which reduces the time by a factor of 2; however, the downsides are an
increase in the statistical variability of the data, which decreases the image signal-tonoise ratio (SNR) and makes the image appear noisy. Methods to reduce acquisition time and/or time per slice are crucial to making MR exam times reasonable, as
described by various methods below.
Multislice Data Acquisition

The average acquisition time per reconstructed image slice in a single-slice spin echo
sequence is clinically unacceptable. However, the average time per slice is significantly
reduced using multislice acquisition methods, where several slices within the tissue
volume are selectively excited in a sequential timing scheme during the TR interval
to fully utilize the dead time waiting for longitudinal recovery in an adjacent slice,
as shown in Figure 13-2. This requires cycling all of the gradients and tuning the RF
excitation pulse many times during the TR interval. The total number of slices that
can be acquired simultaneously is a function of TR, TE, and machine limitations:
Total Number of Slices TR/(TE C),
where C is a constant dependent on the MR equipment capabilities (computer speed;
gradient capabilities; sequence options; additional pulses, e.g., spoiling pulses in
standard SE; use of spatial saturation; and chemical shift, among others). Each slice
and each echo, if multiecho, requires its own k-space repository to store data as it is
acquired. Long TR acquisitions such as proton density and T2-weighted sequences
TR
TE
Repeat
#1
#2
#3
#4
#5
#6
# slices = TR / (TE+C)
FIGURE 13-2 Multislice two-dimensional image acquisition is accomplished by discretely exciting different
slabs of tissue during the TR period; appropriate changes of the RF excitation bandwidth, SSG, PEG, and FEG
parameters are necessary. Because of diffuse excitation profiles, RF irradiation of adjacent slices leads to partial
saturation and loss of contrast. The number of slices (volume) that can be obtained is a function of the TR, TE,
and C, the latter representing the capabilities of the MR system and type of pulse sequence.
452
can produce a greater number of slices over a given volume than T1-weighted
sequences with a short TR. The chief trade-off is a loss of tissue contrast due to crossexcitation of adjacent slices due to nonsquare excitation profiles, causing undesired
proton saturation as explained in Section 13.5 on artifacts.
Data Synthesis
Data synthesis takes advantage of the symmetry and redundant characteristics of
the frequency domain signals in k-space. The acquisition of as little as one-half the
data plus one row of k-space allows the mirroring of complex conjugate data to fill
the remainder of the matrix (Fig. 13-3). In the phase encode direction, half Fourier,
NEX, or phase conjugate symmetry (vendor-specific names) techniques effectively reduce the number of required TR intervals by one-half plus one line, and thus
can reduce the acquisition time by nearly one-half. In the frequency encode direction, fractional echo or read conjugate symmetry refers to reading a fraction of
the echo. While there is no scan time reduction when all the phase encode steps are
acquired, there is a significant echo time reduction, which can reduce motion-related
artifacts, such as dephasing of blood. However, the penalty for either half Fourier or
fractional echo techniques is a reduction in the SNR (caused by a reduced NEX or
Fractional NEX:
Acquired data = matrix + 1 line
Fractional Echo:
minimum TE reduced
kx
peak
min TE
FEG
ky
min TE
peak
ky
Synthesized mirror image data from

opposite quadrants
Mirror
image
kx
Data
extracted
FIGURE 13-3 Fractional NEX and Fractional Echo. Left. Data synthesis uses the redundant characteristics
of the frequency domain. This is an example of phase conjugate symmetry, in which of the PEG views 1
extra are acquired, and the complex conjugate of the data is reflected in the symmetric quadrants. Acquisition
time is thus reduced by approximately 2 ~ 40% , although image noise is increased by approximately 2.
Right: Fractional echo acquisition is performed when only part of the echo is read during the application of the
FEG. Usually, the peak of the echo is centered in the middle of the readout gradient, and the echo signals prior
to the peak are identical mirror images after the peak. With fractional echo, the echo is no longer centered,
and the sampling window is shifted such that only the peak echo and the dephasing part of the echo are
sampled. As the peak of the echo is closer to the RF excitation pulse, TE can be reduced, which can improve
T1 and proton density weighting contrast. A larger number of slices can also be obtained with a shorter TE in
a multislice acquisition (see Fig. 13-2).
453
data sampling in the volume) and the potential for artifacts if the approximations in
the complex conjugation of the signals are not accurate. Other inaccuracies result
from inhomogeneities of the magnetic field, imperfect linear gradient fields, and the
presence of magnetic susceptibility agents in the volume being imaged.
Fast Pulse Sequences

Fast Spin Echo (FSE) techniques use multiple PEG steps in conjunction with multiple 180-degree refocusing RF pulses to produce an echo train length (ETL) with
corresponding digital data acquisitions per TR interval, as illustrated in Figure 13-4.
Multiple k-space rows are filled during each TR equal to the ETL, which is also
the reduction factor for acquisition time. Effective echo time is determined when
the central views in k-space are acquired, which are usually the first echoes, and
subsequent echoes are usually spaced apart via increased PEG strength with the
same echo spacing time. Phase re-ordering optimizes SNR by acquiring the lowfrequency information with the early echoes (lowest amount of T2 decay), and the
high-frequency, peripheral information with late echoes, where the impact on overall image SNR is lower. The FSE technique has the advantage of spin echo image
acquisition, namely immunity from external magnetic field inhomogeneities, with
4, 8, to 16 faster acquisition time. However, each echo experiences different amounts of intrinsic T2 decay, which results in image contrast differences when
compared with conventional spin echo images of similar TR and TE. Lower signal levels in the later echoes produce less SNR, and fewer images can be acquired
in the image volume during the same acquisition. A T2-weighted spin echo image
(TR2,000ms, 256 phase encode steps, one average) requires approximately 8.5min,
while a corresponding FSE with an ETL of 4 (Fig. 13-4) requires about 2.1 min.
TR
90
180
90
RF
SSG
Phase
encode
ordering
PEG
FEG
16 ms
32 ms
48 ms
64 ms
Echo
Effective
TE
Echo train length (ETL) = 4

Effective TE = 16 ms
k-space
FIGURE 13-4 Conventional FSE uses multiple 180-degree refocusing RF pulses per TR interval with incremental changes in the PEG to fill several views in k-space (the ETL). This example illustrates an ETL of four,
with an effective TE equal to 16 ms. Total time of the acquisition is reduced by the ETL factor. The reversed
polarity PEG steps reestablish coherent phase before the next gradient application. Slightly different PEG
strengths are applied to fill the center of k-space first, and then the periphery with later echoes, continuing
until all views are recorded. As shown, data can be mirrored using conjugate symmetry to reduce the overall
time by another factor of two.
454
Longer TR values allow for a greater ETL, which will offset the longer TR in terms of
overall acquisition time, and will also allow more proton density weighting. Specific
FSE sequences for T2 weighting and multiecho FSE are employed with variations
in phase reordering and data acquisition. FSE is also known as turbo spin echo or
RARE (rapid acquisition with refocused echoes).
A Gradient Echo (GE) Acquisition pulse sequence is similar to a standard
spin echo sequence with a readout gradient reversal substituting for the 180-degree
pulse (Fig. 13-5). Repetition of the acquisition sequence occurs for each PEG step
and with each average. With small flip angles and gradient reversals, a considerable reduction in TR and TE is possible for fast image acquisition; however, the
ability to acquire multiple slices is compromised. A PEG rewinder pulse of opposite polarity is applied to maintain phase relationships from pulse to pulse in the
coherent image acquisition. Spoiler gradients are used to eliminate persistent transverse magnetization from stimulated echoes for incoherent GE (see Chapter 12,
Section 12.5).
Acquisition times are calculated in the same way as spin echo; a GE sequence for
a 256 192 image matrix, two averages, and a TR 30 ms, results in an imaging
time equal to 192 2 0.03 s 15.5 s. A conventional spin echo requires 3.84
min for a TR 600 ms. Trade-offs for fast acquisition speed include SNR losses,
magnetic susceptibility artifacts, and less immunity from magnetic field inhomogeneities. There are several acronyms for GE sequences, including GRASS, FISP,
Spoiled GRASS, FLASH, SSFP, etc., depending on the manufacturer of the equipment. Table 13-1 describes a partial list of the different GE sequences and their
method of data acquisition.
Echo Planar Image (EPI) Acquisition is a technique that provides extremely
fast imaging time. Spin Echo (SE-EPI) and Gradient Echo (GE-EPI) are two methods
used for acquiring data, and a third is a hybrid of the two, GRASE (Gradient and Spin
Echo). Single-shot (all of the image information is acquired within 1 TR interval) or
multishot EPI has been implemented with these methods. For single-shot SE-EPI,
TR
RF
SSG
Rewinder gradient
PEG
FEG
Echo
TE
FIGURE 13-5 Coherent GE pulse sequence uses a small flip angle (30 to 40 degrees) RF pulse simultaneous
to the SSG. Phase and frequency encode gradients are applied shortly afterward (with a TE of less than 3 ms
in certain sequences). A PEG rewinder (reverse polarity) reestablishes the phase conditions prior to the next
pulse, simultaneous with the extended FEG duration.
455
TABLE 13-1COMPARISON OF MANUFACTURER-NAMED ACRONYMS FOR
GE SEQUENCES
SEQUENCE
GENERAL ELECTRIC
PHILIPS
SIEMENS
TOSHIBA
Coherent GE
GRASS, FGR FMPGR
FFE
FISP
Field echo
Incoherent GE
(RF spoiled)
SPGR, FSPGR
T1 FFE
Incoherent GE
(Gradient spoiled)
MPGR
Steady-state free
precession
SSFP, DE FGR
T2 FFE
PSIF
SSFP: balanced
sequence / true FISP
FIESTA
Balanced FFE
True FISP
Field echo
FLASH
Field echo
True SSFP
Note: Not all manufacturers are listed in this table, nor are all GE sequences. (Blank areas indicate particular
sequence is not performed (at time of publication).
image acquisition typically begins with a standard 90-degree flip, then a PEG/FEG
gradient application to initiate the acquisition of data in the periphery of the k-space,
followed by a 180-degree echo-producing RF pulse. Immediately after, an oscillating
readout gradient and phase encode gradient blips are continuously applied to stimulate echo formation and rapidly fill k-space in a stepped zig-zag pattern (Fig.13-6).
The effective echo time occurs at a time TE, when the maximum amplitude of
the induced GEs occurs. Acquisition of the data must proceed in a period less than
T2* (around 50 ms), placing high demands on the sampling rate, the gradient coils
(shielded coils are required, with low induced eddy currents), the RF transmitter/
receiver, and RF energy deposition limitations. For GE-EPI, a similar acquisition strategy is implemented but without a 180 degrees refocusing RF pulse, allowing for faster
acquisition time. SE-EPI is generally longer, but better image quality is achieved; on
the other hand, larger RF energy deposition to the patient occurs. EPI acquisition can
Effective TE
90
RF
180
SSG
PEG blips
PEG
FEG
Gradient
Echo
ky
k-space
kx
FIGURE 13-6 Single shot Echo

Planar Spin Echo image (SE-EPI)
acquisition sequence. Data is
deposited in k-space, initially positioned by a simultaneous PEG and
FEG application to locate the initial
row and column position (in this
example, the upper left), followed
by phase encode gradient blips
simultaneous to FEG oscillations,
to fill k-space line by line by introducing 1-row phase changes in
a zig-zag pattern. Image matrix
sizes of 64 64 and 128 64
are common.
456
be preceded with any type of RF pulse, for instance FLAIR (EPI-FLAIR), which will
produce images much faster than the corresponding conventional FLAIR sequence.
The GRASE (Gradient and Spin Echo) sequence combines the initial spin echo
with a series of GEs, followed by an RF rephasing (180 degrees) pulse, and the pattern is repeated until k-space is filled. This hybrid sequence achieves the benefits
of both types of rephasing: the speed of the gradient and the ability of the RF pulse
to compensate for T2* effects, providing significant improvements in image quality
compared to the standard EPI methods. A trade-off is a longer acquisition time (e.g.,
greater than 100 ms) and much greater energy deposition from the multiple 180
degrees RF pulses.
EPI acquisitions typically have poor SNR, low resolution (matrices of 64 64 or
128 64 are typical), and many artifacts, particularly of chemical shift and magnetic
susceptibility origin. Nevertheless, EPI offers real-time snapshot image capability
with 50 ms total acquisition time. EPI is emerging as a clinical tool for studying timedependent physiologic processes and functional imaging. Concerns of safety with
EPI, chiefly related to the rapid switching of gradients and possible nerve stimulation
of the patient, the associated acoustic noise, image artifacts, distortion, and chemical
shift are components that will limit use for many imaging procedures.
Other K-Space Filling Methods

Methods to fill k-space in a nonsequential way can enhance signal, contrast, and
achieve rapid scan times as shown in Figure 13-7. Centric k-space filling has been
discussed with FSE imaging (above), where the lower strength phase encode gradiOuter rows filled last
A Centric filling
Central rows filled first

Outer rows filled last
Outer rows filled first
B Keyhole filling
Central rows filled with

appearance of contrast
Outer rows filled first
C Spiral filling
Equal DT between points

k-space re-binning of spiral
data is required before
image reconstruction
FIGURE 13-7 Alternate methods of filling k-space. A. Centric filling applies the lower strength PEGs first to
maximize signal and contrast from the earliest echoes of a FSE or GE sequence. B. Keyhole filling applies PEGs
of higher strength first to fill the outer portions of k-space, and the central lines are filled only during a certain
part of the sequence, such as with arrival of contrast signal. C. Spiral data acquisition occurs with sinusoidal
oscillation of the X and Y gradients 90 degrees out of phase with each other, with samples beginning in the
center of k-space and spiraling out to the periphery. Interpolating the data into the kx, ky matrix is required in
order to apply 2DFT image reconstruction.
457
ents are applied first, filling the center of k-space when the echoes have their highest
amplitude. This type of filling is also important for fast GE techniques, where the
image contrast and the SNR fall quickly with time from the initial excitation pulse.
Keyhole filling methods fill k-space similarly to centric filling, except the central
lines are filled when important events occur during the sequence, in situations such
as contrast-enhanced angiography. Outer areas of k-space are filled first, and when
gadolinium appears in the imaging volume, the center areas are filled. At the end of
the scan, the outer and central k-space regions are meshed to produce an image with
both good contrast and resolution.
Spiral filling is an alternate method of filling k-space radially, which involves
the simultaneous oscillation of equivalent encoding gradients to sample data points
during echo formation in a spiral, starting at the origin (the center of the k-space)
and spiraling outward to the periphery in the prescribed acquisition plane. The same
contrast mechanisms are available in spiral sequences (e.g., T1, T2, proton density
weighting), and spin or GEs can be obtained. After acquisition of the signals, an additional post-processing step, re-gridding, is necessary to convert the spiral data into
the rectilinear matrix for two-dimensional Fourier transform (2DFT). Spiral scanning
is an efficient method for acquiring data and sampling information in the center of
k-space, where the bulk of image information is contained.
A variant of radial sampling with enhanced filling of the center of k-space is known
generically as blade imaging, and commonly as propeller: Periodically Rotated
Overlapping Parallel Lines with Enhanced Reconstruction, where a rectangular block
of data is acquired and then rotated about the center of k-space. Redundant information concentrated in the center of k-space is used for improvement of SNR or for the
identification of times during the scan in which the patient may have moved, so that
those blocks of data can be processed with a phase-shifting algorithm to eliminate the
movement effect on the data during the reconstruction process and to mitigate motion
artifacts to a great extent. Filling of k-space for this method is shown in Figure 13-8.
Parallel Imaging
Parallel imaging is a technique that fills k-space by using the response of multiple
receive RF coils that are coupled together with independent channels, so that data
can be acquired simultaneously. Specific hardware and software are necessary for the
electronic orchestration of this capability. Typically, 2, 4, 5, 7, 8, 16, 18 (or more)
coils are arranged around the area to be imaged; if a 4-coil configuration is used,
then during each TR period, each coil acquires a view of the data as the acquisition
sequence proceeds. Lines in k-space are defined only after the processing of linear
combinations of the signals that are received by all of the coils. Since 4 views of the
data are acquired per TR interval, scan time can be decreased by a factor of 4 (known
as the reduction factor). However, the acquisition of the signals have gaps, and the
FOV in the phase direction is reduced to one-quarter of its original size. This results
in a known aliasing of the information (a wrapped imagesee section on Artifacts),
that is rectified by using the measured sensitivity profile of each coil to calculate from
where the signal is coming. This sensitivity profile determines the position of the signal based on its amplitude, where the signal near the coil has a higher amplitude than
that farthest away. As a result of the process, commonly known as SENSE (SENSitivity Encodingthere are several acronyms coined by the manufacturers), the image
can be unwrapped and combined with the unwrapped images from each of the other
coils. A simple two-coil example is shown in Figure 13-9 for a breast image application of SENSE, in which improved resolution is desired over reduced scan time.
458
Rectangular filling
Propeller filling
Blades
Motion during
the acquisition
Blades with motion
are processed to
remove motion and
data reconstructed
FIGURE 13-8 The propeller data acquisition compared to a rectangular filling of k-space is shown above.
Instead of acquiring single lines of information to fill k-space consecutively as shown in the upper left and
middle left, a rectangular data acquisition at a specific angle (e.g., 0 degree) is acquired encompassing several
lines of k-space, which represents a blade of information. The partial acquisition is rotated about the center
of k-space at angular increments, which provides a dense sampling of data at the center of k-space and less in
the periphery as shown by the schematic (upper right illustration). If the patient moves during a portion of the
examination (lower left image), the blades in which the motion occurred can be identified, reprocessed, and
the image reconstructed without the motion artifacts (lower right image).
FIGURE 13-9 Parallel imaging with two RF
coils. Top. a single coil acquisition of a breast
MR exam over the full FOV. Middle. Individual
coils with every-other row of k-space being
filled represent FOV, with image overlap
caused by aliasing. Bottom. after SENSE processing, images are combined to deliver twice
the spatial resolution in the left/right (Phase)
direction, with the same imaging time.
Single coil
FOV
Right coil
Left coil
FOV/2
Unwrapped, combined
459
Parallel imaging can be used to either reduce scan times or improve resolution. It
also can be used with most pulse sequences. There are obvious benefits in terms of
scan times and/or resolution, but there is a slight loss of SNR due to the manipulation
of the signals, and chemical shift artifacts (explained in the Artifacts section) may
increase. Patient motion can also cause misalignment between the undersampled
data and the reference scans of the coils.
Three-Dimensional Fourier Transform Image Acquisition

Three-dimensional image acquisition (volume imaging) requires the use of a broadband, nonselective, or slab-selective RF pulse to excite a large volume of protons simultaneously. Two phase gradients are discretely applied in the slice encode
and phase encode directions, prior to the frequency encode (readout) gradient
(Fig. 13-10). The image acquisition time is equal to
TR
# Phase Encode Steps (z-axis) # Phase Encode Steps
(y-axis) # Signal Averages
A three-dimensional Fourier transform (three one-dimensional Fourier transforms)
is applied for each column, row, and depth axis in the image matrix cube. Volumes
obtained can be either isotropic, the same size in all three directions, or anisotropic,
where at least one dimension is different in size. The advantage of the former is
equal resolution in all directions; reformations of images from the volume do not
suffer from degradations of larger sample size from other directions. After the spatial
domain data are obtained, individual two-dimensional slices in any arbitrary plane
are extracted by interpolation of the cube data.
When using a standard TR of 600 ms with one average for a T1-weighted exam, a
128 128 128 cube requires 163 min or about 2.7 h! Obviously, this is unacceptable for standard clinical imaging. GE pulse sequences with TR of 50ms acquire the
same image volume in about 15 min. Another shortcut is with anisotropic voxels, where
the phase encode steps in one dimension are reduced, albeit with a loss of resolution.
A major benefit to isotropic three-dimensional acquisition is the uniform resolution
Anisotropic
Slice Encode
(phase #1)
Isotropic
Frequency Encode
FIGURE 13-10 Three-dimensional image acquisition requires the application of a broadband RF pulse to
excite all of the protons in the volume simultaneously, followed by a phase encode gradient along the slice
encode direction, a phase encode gradient along the phase encode direction, and a frequency encode gradient in the readout direction. Spatial location is decoded sequentially by the Fourier transform along each
encode path, storing intermediate results in the three-dimensional k-space matrix.
460
in all directions when extracting any two-dimensional image from the matrix cube. In
addition, high SNR is achieved compared to a similar two-dimensional image, allowing
reconstruction of very thin slices with good detail (less partial volume averaging) and
high SNR. A downside is the increased probability of motion artifacts and increased
computer hardware requirements for data handling and storage.
13.2 MR Image Characteristics
Spatial Resolution and Contrast Sensitivity

Spatial resolution, contrast sensitivity, and SNR parameters form the basis for evaluating the MR image characteristics. The spatial resolution is dependent on the FOV,
which determines pixel size, the gradient field strength, which determines the FOV,
the receiver coil characteristics (head coil, body coil, and various surface coil designs),
the sampling bandwidth, and the image matrix. Common image matrix sizes are 128
128, 256 128, 256 192, and 256 256, with 512 256, 512 512, and
1,024 512 becoming prevalent. In general, MR provides spatial resolution approximately equivalent to that of CT, with pixel dimensions on the order of 0.5 to 1.0
mm for a high-contrast object and a reasonably large FOV (greater than 250mm).
A 250mm FOV and a 256 256 matrix will have a pixel size on the order of 1
mm. In small FOV acquisitions with high gradient strengths and with surface coil
receivers, the effective pixel size can be smaller than 0.1 to 0.2 mm (of course, with
a limited FOV of 25 to 50 mm. Slice thickness in MRI is usually 5 to 10 mm and
represents the dimension that produces the most partial volume averaging.
Spatial resolution can be improved with higher field strength magnets due to a larger
SNR, which allows thinner slice acquisition, and/or higher sampling rates (smaller pixels) for a given acquisition. However, with higher B0, increased RF absorption, artifact
production, and a lengthening of T1 relaxation occur. The latter decreasesT1 contrast
sensitivity because of increased saturation of the longitudinal m
agnetization.
Contrast sensitivity is the major attribute of MR. The spectacular contrast sensitivity of MR enables the exquisite discrimination of soft tissues and contrast due to blood
flow. This sensitivity is achieved through differences in the T1, T2, proton density, and
flow velocity characteristics. Contrast, which is dependent upon these parameters, is
achieved through the proper application of pulse sequences, as discussed previously.
MR contrast materials, usually susceptibility agents that disrupt the local magnetic field
to enhance T2 decay or provide a relaxation mechanism for shorter T1 recovery time
(e.g., bound water in hydration layers), are becoming important enhancement agents
for differentiation of normal and diseased tissues. The absolute contrast sensitivity of
the MR image is ultimately limited by the SNR and presence of image artifacts.
Signal-to-Noise Ratio, SNR

There are numerous dependencies on the ultimate SNR achievable by the MR system.
The intrinsic signal intensity based on T1, T2, and proton density parameters has
been discussed; to summarize, the TR, TE, and flip angle will have an impact on the
magnitude of the signal generated in the image. While there are many mitigating
factors, a long TR increases the longitudinal magnetization recovery and increases
the SNR; a long TE increases the transverse magnetization decay and reduces the
SNR; a smaller flip angle (reduced from 90 degrees) reduces the SNR. Therefore,
spin echo pulse sequences with large flip angle, long TR, short TE, coarse matrix,
461
large FOV, thick slices, and many averages will generate the best SNR; however, the
resultant image may not be clinically relevant or desirable. While SNR is important,
its noteverything.
For a given pulse sequence (TR, TE, flip angle), the SNR of the MR image is
dependent on a number of variables, as shown in the equation below for a twodimensional image acquisition:
SNR I voxel x, y, z
NEX
f(QF) f(B) f(slice gap) f(reconstruction)
BW
where I is the intrinsic signal intensity based on pulse sequence, voxelx,y,z is the voxel
volume, determined by FOV, image matrix, and slice thickness, NEX is the number
of excitations, determined by the number (or fractional number) of repeated signal
acquisitions into the same voxels. BW is the frequency bandwidth of the RF receiver,
f(QF) is the function of the coil quality factor parameter (tuning the coil), f(B) is the
function of magnetic field strength, B, f(slice gap) is the function of interslice gap
effects, and f(reconstruction) is the function of the reconstruction algorithm.
Other factors in the above equation are explained briefly below.
Voxel Volume
The voxel volume is equal to
Volume
FOVy
FOVx
Slice thickness, z
No. of pixels, x No. of pixels, y
SNR is linearly proportional to the voxel volume. Thus, by reducing the image matrix
size from 256 256 to 256 128 over the same FOV, the effective voxel size
increases by a factor of two, and therefore increases the SNR by a factor of two for the
same image acquisition time (e.g., 256 phase encodes with one average versus 128
phase encodes with two averages).
Signal Averages
Signal averaging (also known as number of excitations, NEX) is achieved by averaging sets of data acquired using an identical pulse sequence (same PEG strength). The
SNR is proportional to the square root of the number of signal averages. A 2-NEX
acquisition requires a doubling (100% increase) of the acquisition time for a 40%
increase in the SNR 2 = 1.4 Doubling the SNR requires 4 NEX. In some cases,
less than 1 average (e.g., or NEX) can be selected. Here, the number of phase
encode steps is reduced by or , and the missing data are synthesized in the
k-space matrix. Imaging time is therefore reduced by a similar amount; however, a
loss of SNR accompanies the shorter imaging times by the same square root factor.
RF Bandwidth
The receiver bandwidth defines the range of frequencies to which the detector is
tuned during the application of the readout gradient. A narrow bandwidth (a narrow spread of frequencies around the center frequency) provides a higher SNR,
1
proportional to BW . A twofold reduction in RF bandwidthfrom 8 to 4 kHz,
462
for instanceincreases the SNR by 1.4 (40% increase). This is mainly related
to the fact that the white noise, which is relatively constant across the bandwidth,
does not change, while the signal distribution changes with bandwidth. In the spatial domain, bandwidth is inversely proportional to the sample dwell time, T to
sample the signal: BW 1/T. Therefore, a narrow bandwidth has a longer dwell
time, which increases the signal height (Fig. 13-11), compared to the shorter dwell
time for the broad bandwidth signal, thus spreading the signal over a larger range
of frequencies. The SNR is reduced by the square root of the dwell time. However,
any decrease in RF bandwidth must be coupled with a decrease in gradient strength
to maintain the sampling across the FOV, which might be unacceptable if chemical shift artifacts are of concern (see Artifacts, below). Narrower bandwidths also
require a longer time for sampling, and therefore affect the minimum TE time that
is possible for an imaging sequence. Clinical situations that can use narrow bandwidths are with T2-weighted images and long TEs that allow the echo to evolve over
an extended period, particularly in situations where fat saturation pulses are used to
reduce the effects of chemical shift in the acquired images. Use of broad bandwidth
settings is necessary when very short TEs are required, such as in fast GE imaging
to reduce the sampling time.
RF Coil Quality Factor

The coil quality factor is an indication of RF coil sensitivity to induced currents in
response to signals emanating from the patient. Coil losses that lead to lower SNR are
caused by patient loading effects and eddy currents, among other factor. Patient loading refers to the electric impedance characteristics of the body, which to a certain extent
acts like an antenna. This effect causes a variation in the magnetic field that is different
Signal
Noise
broad (16 kHz)
DT
(8 kHz)
DT
narrow (4 kHz)
RF bandwidth =
1 / dwell time =
1 / DT
Sample dwell time
DT
FIGURE 13-11 RF Receiver Bandwidth is determined by the FEG strength, the FOV, and sampling rate. This
figure illustrates the spatial domain view of SNR and corresponding sample dwell time. Evolution of the echo
in the broad bandwidth situation occurs rapidly with minimal dwell time, which might be needed in situations
where very short TE is required, even though the SNR is reduced. On the other hand, in T2 weighted images
requiring a long TE, narrow bandwidth can improve SNR.
463
for each patient, and must be measured and corrected for. Consequently, tuning the
receiver coil to the resonance frequency is mandatory before image acquisition. Eddy
currents are signals that are opposite of the induced current produced by transverse
magnetization in the RF coil, and reduce the overall signal. Quadrature coils increase the
SNR as two coils are used in the reception of the signal; phased array coils increase the
SNR even more when the data from several coils are added together (see Parallel Imaging, Section 13.1). The proximity of the receiver coil to the volume of interest affects the
coil quality factor, but there are trade-offs with image uniformity. Positioning of the coil
with respect to the direction of the main magnetic field is also an issue that occurs with
air core (horizontal B0) to solid core (vertical B0) magnets. Body receiver coils positioned
in the bore of the magnet have a moderate quality factor, whereas surface coils have a
high quality factor. With the body coil, the signal is relatively uniform across the FOV;
however, with surface coils, the signal falls off abruptly near the edges of the field, limiting the useful imaging depth and resulting in nonuniform brightness across the image.
Magnetic Field Strength

Magnetic field strength influences the SNR of the image by a factor of B1.0 to B1.5.
Thus, one would expect a three- to fivefold improvement in SNR with a 1.5 T magnet
over a 0.5 T magnet. Although the gains in the SNR are real, other considerations
mitigate the SNR improvement in the clinical environment, including longer T1
relaxation times and greater RF absorption, as discussed previously.
Cross-Excitation
Cross-excitation occurs from the nonrectangular RF excitation profiles in the spatial
domain and the resultant overlap of adjacent slices in multislice image acquisition
sequences. This saturates the protons and reduces contrast and the contrast-to-noise
ratio. To avoid cross-excitation, interslice gaps or interleaving procedures are necessary (see Artifacts section, below).
Image Acquisition and Reconstruction Algorithms

Image acquisition and reconstruction algorithms have a profound effect on SNR. The
various acquisition/reconstruction methods that have been used in the past and those
used today are, in order of increasing SNR, point acquisition methods, line acquisition
methods, two-dimensional Fourier transform acquisition methods, and three-dimensional Fourier transform volume acquisition methods. In each of these techniques,
the volume of tissue that is excited is the major contributing factor to improving the
SNR and image quality. Reconstruction filters and image processing algorithms will
also affect the SNR. High-pass filtration methods that increase edge definition will
generally decrease the SNR, while low-pass filtration methods that smooth the image
data will generally increase the SNR at the cost of reduced resolution.
Summary, Image Quality

The best possible image quality is always desirable, but not always achievable because
of the trade-off between SNR, scan speed, and spatial resolution. To increase one of
these three components of image quality involves the consideration of reducing one
or both of the other two. It is thus a balancing act that is chosen by the operator, the
protocol, and the patient in order to acquire images with the best diagnostic yield.
464
MR parameters that may be changed include TR, TE, TI, ETL, Matrix Size, Slice
Thickness, Field of view, and NEX. Working with these parameters in the optimization of acquisition protocols to achieve high image quality is essential.
13.3 Signal from Flow

The appearance of moving fluid (vascular and cerebrospinal fluid [CSF]) in MR images
is complicated by many factors, including flow velocity, vessel orientation, laminar
versus turbulent flow patterns, pulse sequences, and image acquisition modes. Flowrelated mechanisms combine with image acquisition parameters to alter contrast.
Signal due to flow covers the entire gray scale of MR signal intensities, from black
blood to bright blood levels, and flow can be a source of artifacts. The signal from
flow can also be exploited to produce MR angiographic images.
Low signal intensities (flow voids) are often a result of high-velocity signal loss
(HVSL), in which protons in the flowing blood move out of the slice during echo reformation, causing a lower signal. Flow turbulence can also cause flow voids, by causing
a dephasing of protons in the blood with a resulting loss of the tissue magetization in
the area of turbulence. With HVSL, the amount of signal loss depends on the velocity
of the moving fluid. Pulse sequences to produce black blood in images can be very
useful in cardiac and vascular imaging. A typical black blood pulse sequence uses a
double inversion recovery method, whereby a nonselective 180-degree RF pulse
is initially applied, inverting all protons in the body, and is followed by a selective
180-degree RF pulse that restores the magnetization in the selected slice. During the
inversion time, blood outside of the excited slice with inverted protons flows into the
slice, producing no signal; therefore, the blood appears dark.
Flow-Related Enhancement
Flow-related enhancement is a process that causes increased signal intensity due to
flowing protons; it occurs during imaging of a volume of tissues. Even-echo rephasing is a phenomenon that causes flow to exhibit increased signal on even echoes in
a multiple-echo image acquisition. Flowing protons that experience two subsequent
180-degree pulses (even echoes) generate higher signal intensity due to a constructive rephasing of protons during echo formation. This effect is prominent in slow
laminar flow (e.g., veins show up as bright structures on even-echo images).
Flow enhancement in GE images is pronounced for both venous and arterial
structures, as well as CSF. The high intensity is caused by the wash-in (between
subsequent RF excitations) of fully unsaturated protons into a volume of partially
saturated protons due to the short TR used with gradient imaging. During the next
excitation, the signal amplitude resulting from the moving unsaturated protons is
about 10 times greater than that of the nonmoving saturated protons. With GE techniques, the degree of enhancement depends on the velocity of the blood, the slice
or volume thickness, and the TR. As blood velocity increases, unsaturated blood
exhibits the greatest signal. Similarly, a thinner slice or decreased repetition time
results in higher flow enhancement. In arterial imaging of high-velocity flow, it is
possible to have bright blood throughout the imaging volume of a three-dimensional
acquisition if unsaturated blood can penetrate into the volume prior to experiencing
an RF pulse.
Signal from blood is dependent on the relative saturation of the surrounding
tissues and the incoming blood flow in the vasculature. In a multislice volume,

Imaging volume
Unsaturated spins: high signal

Flow-Related Enhancement
Presaturation
RF pulses outside
imaging volume
465
FIGURE 13-12 The repeated RF excitation within

an imaging volume produces partial saturation of
the tissue magnetization (top figure, gray area).
Unsaturated protons flowing into the volume generate a large signal difference that is bright relative to
the surrounding tissues. Bright blood effects can be
reduced by applying pre-saturation RF pulses adjacent
to the imaging volume, so that protons in inflowing
blood will have a similar partial saturation (bottom
figure; note no blood signal).
Pre-saturated spins: equal signal

Flow presaturation
repeated excitation of the tissues and blood causes a partial saturation of the protons,
dependent on the T1 characteristics and the TR of the pulse sequence. Blood outside of the imaged volume does not interact with the RF excitations, and therefore
these unsaturated protons may enter the imaged volume and produce a large signal
compared to the blood within the volume. This is known as flow-related enhancement. As the pulse sequence continues, the unsaturated blood becomes partially
saturated and the protons of the blood produce a similar signal to the tissues in the
inner slices of the volume (Fig. 13-12). In some situations, flow-related enhancement is undesirable and is eliminated with the use of presaturation pulses applied
to volumes just above and below the imaging volume. These same saturation pulses
are also helpful in reducing motion artifacts caused by adjacent tissues outside the
imaging volume.
MR Angiography
Exploitation of blood flow enhancement is the basis for MRA. Two techniques to create
images of vascular anatomy include time-of-flight and phase contrast a ngiography.
Time-of-Flight Angiography
The time-of-flight technique relies on the tagging of blood in one region of the body
and detecting it in another. This differentiates moving blood from the surround stationary tissues. Tagging is accomplished by proton saturation, inversion, or relaxation to
change the longitudinal magnetization of moving blood. The penetration of the tagged
blood into a volume depends on the T1, velocity, and direction of the blood. Since the
detectable range is limited by the eventual saturation of the tagged blood, long vessels
are difficult to visualize simultaneously in a three-dimensional volume. For these reasons, a two-dimensional stack of slices is typically acquired, where even slowly moving blood can penetrate the region of RF excitation in thin slices (Fig.13-13). Each
slice is acquired separately, and blood moving in one direction (north or south, e.g.,
arteries versus veins) can be selected by delivering a presaturation pulse on an adjacent slab superior or inferior to the slab of data acquisition. Thin slices are also helpful in preserving resolution of the flow pattern. Often used for the two-dimensional
466
FIGURE 13-13 The time of flight MRA acquisition collects each slice separately with a sequence to enhance
blood flow. Exploitation of blood flow is achieved by detecting unsaturated protons moving into the volume,
producing a bright signal. A coherent GE image acquisition pulse sequence is shown, TR 24 ms, TE 3.1
ms, Flip Angle 20 degrees. Every 10th image in the stack is displayed above, from left to right and top to
bottom.
image acquisition is a GRASS or FISP GE technique that produces relatively poor

anatomic contrast, yet provides a high-contrast bright blood signal. Magnetization
transfer contrast sequences (see below) are also employed to increase the contrast of
the signals due to blood by reducing the background anatomic contrast.
Two-dimensional TOF MRA images are obtained by projecting the content of the
stack of slices at a specific angle through the volume. A maximum intensity projection (MIP) algorithm detects the largest signal along a given ray through the volume
and places this value in the image (Fig. 13-14). The superimposition of residual
FIGURE 13-14 A simple illustration shows
how the MIP algorithm extracts the highest
(maximum) signals in the two-dimensional
stack of images along a specific direction in
the volume, and produces projection images
with maximum intensity variations as a function of angle.
Projections are cast through the image stack (volume)

The maximum signal along each line is projected
MIP images
Projection angles
Volume of 2D Images
467
stationary anatomy often requires further data manipulation to suppress undesirable

signals. This is achieved in a variety of ways, the simplest of which is setting a window threshold. Another method is to acquire a dataset without contrast, and subtract
the noncontrast MIP from the contrast MIP to reduce background signals. Clinical MRA images show the three-dimensional characteristics of the vascular anatomy
from several angles around the volume stack (Fig. 13-15) with some residual signals
from the stationary anatomy. Time-of-flight angiography often produces variation in
vessel intensity dependent on orientation with respect to the image plane, a situation
that is less than optimal.
Phase Contrast Angiography

Phase contrast imaging relies on the phase change that occurs in moving protons
such as blood. One method of inducing a phase change is dependent on the application of a bipolar gradient (one gradient with positive polarity followed by a second
gradient with negative polarity, separated by a delay time T). In a second acquisition of the same view of the data (same PEG), the polarity of the bipolar gradients is
reversed, and moving protons are encoded with negative phase, while the stationary
protons exhibit no phase change (Fig. 13-16). Subtracting the second excitation from
the first cancels the magnetization due to stationary protons but enhances magnetization due to moving protons. Alternating the bipolar gradient polarity for each
subsequent excitation during the acquisition provides phase contrast image information. The degree of phase shift is directly related to the velocity encoding (VENC)
time, T, between the positive and negative lobes of the bipolar gradients and the
velocity of the protons within the excited volume. Proper selection of the VENC time
Data volume
2D Projection
Angiograms from MIP
FIGURE 13-15 A volume stack of bright blood images (left) is used with MIP processing to create a series
of projection angiograms at regular intervals; the three-dimensional perspective is appreciated in a stack view,
with virtual rotation of the vasculature.
468

bipolar gradient encoding
T
T
Excitation
#1
Excitation
#2
Excitation #1 - Excitation #2 = net residual phase
Stationary spins:
net phase shift = 0
Moving spins, low

velocity, forward direction:
net phase shift is small;
Backward direction, high
flow, net phase shift is
large, opposite
FIGURE 13-16 Phase Contrast Angiography uses consecutive excitations that have a bipolar gradient
encoding with the polarity reversed between the first and second excitation, as shown in the top row. Magnetization vectors (lower two rows) illustrate the effect of the bipolar gradients on stationary and moving
spins for the first and second excitations. Subtracting the two will cancel stationary tissue magnetization and
enhance phase differences caused by the velocity of moving blood.
is necessary to avoid phase wrap error (exceeding 180-degree phase change) and to
ensure an optimal phase shift range for the velocities encountered. Intensity variations are dependent on the amount of phase shift, where the brightest pixel values
represent the largest forward (or backward) velocity, a mid-scale value represents 0
velocity, and the darkest pixel values represent the largest backward (or forward)
velocity. Figure 13-17 shows a representative magnitude and phase contrast image of
the cardiac vasculature. Unlike the time-of-flight methods, the phase contrast image
Magnitude Image
Phase Image
FIGURE 13-17 Magnitude (left) and phase (right) images provide contrast of flowing blood. Magnitude
images are sensitive to flow, but not to direction; phase images provide direction and velocity information.
The blood flow from the heart shows forward flow in the ascending aorta (dark area) and forward flow in the
descending aorta at this point in the heart cycle for the phase image. Some bright flow patterns in the ascending aorta represent backward flow to the coronary arteries. Grayscale amplitude is proportional to velocity,
where intermediate grayscale is 0 velocity.
469
is inherently quantitative, and when calibrated carefully, provides an estimate of the

mean blood flow velocity and direction. Two- and three-dimensional phase contrast
image acquisitions for MRA are possible.
Gradient Moment Nulling

In spin echo and gradient recalled echo imaging, the slice select and readout gradients are
balanced, so that the uniform dephasing with the initial gradient application is rephased
by an opposite polarity gradient of equal area. However, when moving protons are subjected to the gradients, the amount of phase dispersion is not compensated (Fig. 13-18).
This phase dispersal can cause ghosting (faint, displaced copies of the anatomy) in images.
It is possible, however, to rephase the protons by a gradient moment nulling technique.
With constant velocity flow (first-order motion), all protons can be rephased using the
application of a gradient triplet. In this technique, an initial positive gradient of unit area
is followed by a negative gradient of twice the area, which creates phase changes that are
compensated by a third positive gradient of unit area. The velocity compensated gradient (right graph in Fig. 13-18) depicts the evolution of the proton phase back to zero
for both stationary and moving protons. Note that the overall applied gradient has a net
area of zeroequal to the sum of the positive and negative areas. Higher order corrections such as second- or third-order moment nulling to correct for acceleration and other
motions are possible, but these techniques require more complicated gradient switching.
Gradient moment nulling can be applied to both the slice select and readout gradients to
correct for problems such as motion ghosting as elicited by CSF pulsatile flow.
13.3 Perfusion and Diffusion Contrast Imaging

Perfusion of tissues via the capillary bed permits the delivery of oxygen and nutrients
to the cells and removal of waste (e.g., CO2) from the cells. Conventional perfusion
measurements are based on the uptake and wash-out of radioactive tracers or other
exogenous tracers that can be quantified from image sequences using well-characterized
imaging equipment and calibration procedures. For MR perfusion images, exogenous
RF pulse
Net area = 0
Slice
Select
Relative
phase
change
Stationary
spins
Moving
spins
Stationary
and moving
spins
FIGURE 13-18 Left. Phase dispersion of stationary and moving spins under the influence of an applied
gradient (no flow compensation) as the gradient is inverted is shown. The stationary spins return to the original
phase state, whereas the moving spins do not. Right. Gradient moment nulling of first order linear velocity
(flow compensation) requires a doubling of the negative gradient amplitude followed by a positive gradient
such that the total summed area is equal to zero. This will return both the stationary spins and the moving
spins back to their original phase state.
470
and endogenous tracer methods are used. Freely diffusible tracers using nuclei such
as 2H (deuterium), 3He, 17O, and 19F are employed in experimental procedures to
produce differential contrast in the tissues. More clinically relevant are intravascular
blood pool agents such as gadoliniumdiethylenetriaminepentaacetic acid, which
modify the relaxation of protons in the blood in addition to producing a shorter T2*.
This produces signal changes that can be visualized in pre and post contrast images
(Fig. 13-19). Endogenous tracer methods do not use externally added agents, but
instead depend on the ability to generate contrast from specific excitation or diffusion
mechanisms. For example, labeling of inflowing protons (black blood perfusion)
uses protons in the blood as the contrast agent. Tagged (labeled) protons outside of
the imaging volume perfuse into the tissues, resulting in a drop in signal intensity, a
time course of events that can be monitored by quantitative measurements.
Functional MRI (fMRI) is based on the increase in blood flow to the local vasculature that accompanies neural activity in the specific areas of the brain, resulting in a
local reduction of deoxyhemoglobin because the increase in blood flow occurs without
an increase in oxygen extraction. As deoxyhemoglobin is a paramagnetic agent, it alters
the T2*-weighted MRI image signal. Thus, this endogenous contrast enhancing agent
serves as the signal for fMRI. Area voxels (represented by x-y coordinates and z slice
thickness) of high metabolic activity resulting from a task-induced stimulus produce a
correlated signal for Blood Oxygen Level Dependent (BOLD) acquisition techniques. A
BOLD sequence produces multiple T2*-weighted images of the head before the application of the stimulus. The patient is repeatedly subjected to the stimulus and multiple
BOLD images are acquired. Because the BOLD sequence produces images that are
highly dependent on blood oxygen levels, areas of high metabolic activity will demonstrate a change in signal when the prestimulus image data set is subtracted, voxel
by voxel, from post-stimulus image data set. Voxel locations defined by significant
FIGURE 13-19 Pre (top row) and post (bottom row) gadolinium contrast T1-weighted MR axial images
of the brain illustrate the signal change that occurs with the appearance of gadolinium by shortening the T1
time of the perfused tissues.
471
signal changes indicate regions of the brain activated by a specific task. Stimuli in fMRI
experiments can be physical (finger movement), sensory (light flashes or sounds), or
cognitive (repetition of good or bad word sequences, complex problem solving),
among others. To improve the SNR in the fMRI images, a stimulus is typically applied
in a repetitive, periodic sequence, and BOLD images are acquired continuously, tagged
with the timing of the stimulus. Regions in the brain that demonstrate time-dependent
activity and correlate with the time-dependent application of the stimulus are statistically analyzed, and coded using a color scale, while voxels that do not show a
significant intensity change are not colored. The resultant color map is overlaid onto a
grayscale image of the brain for anatomic reference, as shown in Figure 13-20.
High-speed imaging and T2* weighting necessary for fMRI is typically achieved
with EPI acquisition techniques that can be acquired in as little as 50 ms for a 64
64 acquisition matrix. Gradient Recalled Echo acquisitions using standard sequences
are also employed with multiple contiguous slices (e.g., 16 slices, slice thickness 5
mm, TR 3 s, TE 60 ms, 90-degree flip angle) at 1.5 T, with 25 to 30 complete
head acquisitions. The latter acquisition techniques provide better spatial resolution
but rely on very cooperative subjects and a much longer exam time.
Diffusion Weighted Imaging

Diffusion relates to the random motion of water molecules in tissues. Interaction of
the local cellular structure with the movement of water molecules produces anisotropic, directionally dependent diffusion (e.g., in the white matter of brain tissues).
Bilateral finger tapping paradigm
r=0.729
r=0.668
FIGURE 13-20 Functional MR image bilateral finger tapping paradigm shows the areas of the brain activated by this repeated activity. The paradigm was a right finger tap alternated by a left finger tap (time
sequence on the right side of the figure) and the correlated BOLD signals (black traces) derived from the echo
planar image sequence. A voxel-by-voxel correlation of the periodic stimulus and MR signal is performed, and
when exceeding a correlation threshold, a color overlay is added to the grayscale image. In this example, red
indicates the right finger tap that excites the left motor cortex, which appears on the right side of the image,
and blue the left finger tap. (Figure compliments of MH Buonocore, MD, PhD University of California Davis.)
472
FIGURE 13-21 The basic elements of a DWI

pulse sequence are shown. The diffusion weighting gradients are of amplitude G, duration of the
gradients and time between gradients is .
90
RF
SSG
180
DWI
gradients
G
D
PEG
FEG
Signal
Freely diffusing water
Restricted diffusion
Normal Tissue
Ischemic Injury / Stroke
Diffusion sequences use strong MR gradients applied symmetrically about the refocusing pulse to produce signal differences based on the mobility and directionality of
water diffusion, as shown in Figure 13-21. Tissues with more water mobility (normal)
have a greater signal loss than those of lesser water mobility (injury) under the influence of the diffusion weighted imaging (DWI) gradients.
The in vivo structural integrity of certain tissues (healthy, diseased, or injured) can
be measured by the use of DWI, in which water diffusion characteristics are determined through the generation of apparent diffusion coefficient (ADC) maps. This
requires two or more acquisitions with different DWI parameters. A low ADC corresponds to high signal intensity on a calculated image, which represents restricted
diffusion (Fig.13-22). ADC maps of the brain and the spinal cord have shown prom-
FIGURE 13-22 Left. Diffusion weighted image. Right. Calculated ADC image, showing an area of increased
brightness related to restricted mobility of water molecules.
473
ise in predicting and evaluating pathophysiology before it is visible on conventional

T1- or T2-weighted images. DWI is also a sensitive indicator for early detection
of ischemic injury. Areas of acute stroke show a drastic reduction in the diffusion
coefficient compared with nonischemic tissues. Diagnosis of multiple sclerosis is a
potential application, based on the measurements of the diffusion coefficients in threedimensional space.
Various acquisition techniques are used to generate diffusion-weighted contrast.
Standard spin echo and EPI pulse sequences with applied diffusion gradients of high
strength are used. Challenges for DWI are the extreme sensitivity to motion of the
head and brain, which is chiefly caused by the large pulsed gradients required for the
diffusion preparation. Eddy currents are also an issue, which reduce the effectiveness
of the gradient fields, so compensated gradient coils are necessary. Several strategies
have been devised to overcome the motion sensitivity problem, including common
electrocardiographic gating and motion compensation methods.
13.4 Magnetization Transfer Contrast

Magnetization transfer contrast is the result of selective observation of the interaction
between protons in free water molecules and protons contained in the macromolecules
of a protein. Magnetization exchange occurs between the two proton groups because
of coupling or chemical exchange. Because the protons exist in slightly different magnetic environments, the selective saturation of the protons in the macromolecule can
be excited separately from the bulk water by using narrow-band RF pulses (because the
Larmor frequencies are different). A transfer of the magnetization from the protons in
the macromolecule partially saturates the protons in bulk water, even though these protons have not experienced an RF excitation pulse (Fig. 13-23). Reduced signal from the
Hydration Layer
Macromolecule
Protons bound to
macromolecule
H H
H H
O
H H
H H
O
O
O
H H
OH H H
H H
OH H
H H
O
O
O
H H H HO
H H
O O
H H
H H H H
O
O
O
O
H H
H H
NH3 O
O
H H H H
H HO
O
H H
O H H
H H H H
O
H H O
O
O
O H H
O
H H
H H
H H
O
O H H
H H
O
H H O
O
O
RF Excitation
(off-resonance pulse)
~1.5 kHz shift
Bulk Water
Protons of
bulk water
Frequency Spectrum
FIGURE 13-23 Magnetization transfer contrast is implemented with an off-resonance RF pulse of about
1,500 Hz from the Larmor frequency. Excitation of hydrogen atoms on macromolecules is transferred via the
hydration layer to adjacent free-water hydrogen atoms. A partial saturation of the tissues reduces the signals that would otherwise compete with signals from blood flow, making this useful for time-of-flight MRA.
474
adjacent free water protons by the saturation label affects only those protons having a
chemical exchange with the macromolecules and improves local image contrast in many
situations by decreasing the otherwise large signal generated by the protons in the bulk
water. This technique is used for anatomic MRI of the heart, the eye, multiple sclerosis,
knee cartilage, and general MRA. Tissue characterization is also possible, because the
image contrast in part is caused by the surface chemistry of the macromolecule and the
tissue-specific factors that affect the magnetization transfer characteristics.
Magnetization transfer contrast pulse sequences are often used in conjunction
with MRA time-of-flight methods. Hydrogen atoms constitute a large fraction of macromolecules in proteins, are tightly bound to these macromolecules, and have a very
short T2 decay with a broad range of resonance frequencies compared to protons in
free water. Selective excitation of these protons is achieved with an off-resonance RF
pulse of approximately 1,500 Hz from the Larmor frequency, causing their saturation. The protons in the hydration layer bound to these molecules are affected by
the magnetization and become partially saturated themselves. MR signals from these
tissues are suppressed, with an impact of reducing the contrast variation of the anatomy. As a result, the differential contrast of the flow-enhanced signals is increased,
with overall better image quality angiographic sequence.
13.5 MR Artifacts
Artifacts manifest as positive or negative signal intensities that do not accurately represent the imaged anatomy. Although some artifacts are relatively insignificant and
are easily identified, others can limit the diagnostic potential of the exam by obscuring or mimicking pathologic processes or anatomy. One must realize the impact
of MR acquisition protocols and understand the etiology of artifact production to
exploit the information they convey.
To minimize the impact of MR artifacts, a working knowledge of MR physics as
well as image acquisition techniques is required. On the one hand, there are many
variables and options available that complicate the decision-making process for MR
image acquisition. On the other, the wealth of choices enhances the goal of achieving
diagnostically accurate images. MR artifacts are classified into three broad areas
those based on the machine, on the patient, and on signal processing.
Machine-Dependent Artifacts
Magnetic field inhomogeneities are either global or focal field perturbations that lead
to the mismapping of tissues within the image, and cause more rapid T2 relaxation.
Distortion or misplacement of anatomy occurs when the magnetic field is not completely homogeneous. Proper site planning, self-shielded magnets, automatic shimming, and preventive maintenance procedures help to reduce inhomogeneities.
Focal field inhomogeneities arise from many causes. Ferromagnetic objects in or
on the patient (e.g., makeup, metallic implants, prostheses, surgical clips, dentures)
produce field distortions and cause protons to precess at frequencies different from
the Larmor frequency in the local area. Incorrect proton mapping, displacement, and
appearance as a signal void with a peripherally enhanced rim of increased signal are
common findings. Geometric distortion of surrounding tissue is also usually evident. Even nonferromagnetic conducting materials (e.g., aluminum) produce field
distortions that disturb the local magnetic environment. Partial compensation by
the spin echo (180-degree RF) pulse sequence reduces these artifacts; on the other
475
hand, the gradient-refocused echo sequence accentuates distortions, since the protons
always experience the same direction of the focal magnetic inhomogeneities within
the patient.
Susceptibility Artifacts
Magnetic susceptibility is the ratio of the induced internal magnetization in a tissue
to the external magnetic field. As long as the magnetic susceptibility of the tissues
being imaged is relatively unchanged across the field of view, then the magnetic field
will remain uniform. Any drastic changes in the magnetic susceptibility will distort
the magnetic field. The most common susceptibility changes occur at tissue-air interfaces (e.g., lungs and sinuses), which cause a signal loss due to more rapid dephasing
(T2*) at the tissue-air interface (Fig. 13-24). Any metal (ferrous or not) may have
a significant effect on the adjacent local tissues due to changes in susceptibility and
the resultant magnetic field distortions. Paramagnetic agents exhibit a weak magnetization and increase the local magnetic field causing an artifactual reduction in the
surrounding T2* relaxation.
Magnetic susceptibility can be quite helpful in some diagnoses. Most notable is the
ability to diagnose the age of a hemorrhage based on the signal characteristics of the
blood degradation products, which are different in the acute, subacute, and chronic
phases. Some of the iron-containing compounds (deoxyhemoglobin, methemoglobin,
hemosiderin, and ferritin) can dramatically shorten T1 and T2 relaxation of nearby protons. The amount of associated free water, the type and structure of the iron-containing
molecules, the distribution (intracellular versus extracellular), and the magnetic field
strength all influence the degree of relaxation effect that may be seen. For example,
in the acute stage, T2 shortening occurs due to the paramagnetic susceptibility of the
organized deoxyhemoglobin in the local area, without any large effect on the T1 relaxation time. When red blood cells lyse during the subacute stage, the hemoglobin is
altered into methemoglobin, and spin-lattice relaxation is enhanced with the formation
of a hydration layer, which shortens T1 relaxation, leading to a much stronger signal
FIGURE 13-24 Susceptibility artifacts due to dental fillings are shown in the same axial image slice. Left.
Axial T2-weighted fast spin echo image illustrates significant suppression of susceptibility artifacts with
180-degree refocusing pulse. Right. Axial T2*-weighted gradient echo image illustrates significant image void
exacerbated by the gradient echo, where external inhomogeneities are not canceled in the reformed echo.
476
on T1-weighted images. Increased signal intensity on T1-weighted images not found

in the acute stage of hemorrhage identifies the subacute stage. In the chronic stage,
hemosiderin, found in the phagocytic cells in sites of previous hemorrhage, disrupts
the local magnetic homogeneity, causes loss of signal intensity, and leads to signal void,
producing a characteristic dark rim around the hemorrhage site.
Gadolinium-based contrast agents (paramagnetic characteristics shorten T2 and
hydration layer interactions shorten T1) are widely used in MRI. Tissues that uptake
gadolinium contrast agents exhibit shortened T1 relaxation and demonstrate increased
signal on T1-weighted images. Although focal inhomogeneities are generally considered problematic, there are certain physiologic and anatomic manifestations that can
be identified and diagnostic information obtained.
Gradient Field Artifacts

Magnetic field gradients spatially encode the location of the signals emanating from
excited protons within the volume being imaged. Proper reconstruction requires linear, matched, and properly sequenced gradients. The slice select gradient defines the
volume (slice). Phase and frequency encoding gradients provide the spatial information in the other two dimensions.
Since the reconstruction algorithm assumes ideal, linear gradients, any deviation
or temporal instability will be represented as a distortion. Gradient strength has a
tendency to fall off at the periphery of the FOV. Consequently, anatomic compression occurs, especially pronounced on coronal and sagittal images having a large
FOV, typically greater than 35 cm (Fig. 13-25). Minimizing the spatial distortion
entails either reducing the FOV by lowering the gradient field strength or by holding the gradient field strength and number of samples constant while decreasing the
frequency bandwidth. Of course, gradient calibration is part of a continuous quality
control (QC) checklist, and geometric accuracy must be periodically verified.
Anatomic proportions may simulate abnormalities, so verification of pixel dimensions in the PEG and FEG directions are necessary. If the strength of the FEG and the
strength of the largest PEG are different, the height or width of the pixels can become distorted and produce inaccurate measurements. Ideally, the phase and frequency encoding gradients should be assigned to the smaller and larger dimensions of the object,
respectively, to preserve spatial resolution while limiting the number of phase encoding
steps. In practice, this is not always possible, because motion artifacts or high-intensity
Accurate
geometry
Ideal
Actual
Distorted
FOV
FIGURE 13-25 Gradient nonlinearity causes image distortions by mis-mapping anatomy. In the above
examples, the strength of the gradient at the periphery is less than the ideal (orange line versus black line). This
results in a compression of the imaged anatomy, with inaccurate geometry (images with orange border). For
comparison, images acquired with linear corrections are shown above.
477
signals that need to be displaced away from important areas of interest after an initial
scan might require swapping the frequency and phase encode gradient directions.
RF Coil Artifacts
RF surface coils produce variations in uniformity across the image caused by RF
excitation variability, attenuation, mismatching, and sensitivity falloff with distance.
Proximal to the surface coil, receive signals are intense, and with distance, signal
intensity is attenuated, resulting in grayscale shading and loss of brightness in the
image. Nonuniform image intensities are the all-too-frequent result. Also, compensation for the disturbance of the magnetic field by the patient is typically compensated
by an automatic shimming calibration. When this is not performed, or performed
inadequately, a significant negative impact on image quality occurs. Examples of variable response are shown in Figure 13-26.
Other common artifacts from RF coils occur with RF quadrature coils (coils that
simultaneously measure the signal from orthogonal views) that have two separate amplifier and gain controls. If the amplifiers are imbalanced, a bright spot in the center of the
image, known as a center point artifact, arises as a 0 frequency direct current offset.
Variations in gain between the quadrature coils can cause ghosting of objects diagonally
in the image. The bottom line for all RF coils is the need for continuous measurement
and consistent calibration of their response, so that artifacts are minimized.
RF Artifacts
RF pulses and precessional frequencies of MRI instruments occupy the same frequencies of common RF sources, such as TV and radio broadcasts, electric motors,
fluorescent lights, and computers. Stray RF signals that propagate to the MRI antenna
can produce various artifacts in the image. Narrow-band noise creates noise patterns perpendicular to the frequency encoding direction. The exact position and
spatial extent depends on the resonant frequency of the imager, applied gradient
field strength, and bandwidth of the noise. A narrow band pattern of black/white
Coil close to skin
Inadequate shimming for fat saturation
FIGURE 13-26 Signal intensity variations occur when surface RF receive coils are too close to the skin, as exemplified by the MR breast image on the left. With inadequate shimming calibration, saturation pulses for adipose
tissue in the breast is uneven, causing a significant variation in the uniformity of the reconstructed image. From
Hendrick RE, Breast MRI: fundamentals and technical aspects. New York, NY: Springer, 2007. By permission.
478
alternating noise produces a zipper artifact. Broadband RF noise disrupts the image
over a much larger area of the reconstructed image with diffuse, contrast-reducing
herringbone artifacts. Appropriate site planning and the use of properly installed RF
shielding materials (e.g., a Faraday cage) reduce stray RF interference to an acceptably low level. An example RF zipper artifact is shown in Figure 13-44.
RF energy received by adjacent slices during a multislice acquisition excite and
saturate protons in adjacent slices, chiefly due to RF pulses without sharp off/on/
off transitions. This is known as cross-excitation. On T2-weighted images, the sliceto-slice interference degrades the SNR; on T1-weighted images, the extra partial
saturation reduces image contrast by reducing longitudinal recovery during the TR
interval. A typical truncated sinc RF profile and overlap areas in adjacent slices are
shown in Figure 13-27. Interslice gaps reduce the overlap of the profile tails, and
pseudo-rectangular RF pulse profiles reduce the spatial extent of the tails. Important
anatomic findings could exist within the gaps, so slice interleaving is a technique to
mitigate cross-excitation by reordering slices into two groups with gaps. During the
first half of the TR, the first slices are acquired (slices 1 to 5), followed by the second
group of slices that are positioned in the gap of the first group (slices 6 to 10). This
method reduces cross-excitation by separating the adjacent slice excitations in time.
The most effective method is to acquire two independent sets of gapped multislice
images, but the image time is doubled. The most appropriate solution is to devise RF
pulses that approximate a rectangular profile; however, the additional time necessary
for producing such an RF pulse can be prohibitive.
K-Space Errors
Errors in k-space encoding affect all areas of the reconstructed image, and cause
the artifactual superimposition of wave patterns across the FOV. Each individual
pixel value in k-space contributes to all pixel values in image space as a frequency
harmonic with a signal amplitude. One bad pixel introduces a significant artifact,
rendering the image suboptimal, as shown in Figure 13-28.
FIGURE 13-27 Top. Poor pulse profiles are caused by truncated RF pulses,
and resulting profile overlap causes
unwanted partial saturation in adjacent
slices, with a loss of SNR and CNR. Optimized pulses are produced by considering the tradeoff of pulse duration versus
excitation profile. Bottom. Reduction of
cross-excitation is achieved with interslice gaps, but anatomy at the gap location might be missed. An interleaving
technique acquires the first half of the
images with an interslice gap, and the
second half of the images are positioned
in the gaps of the first images. The separation in time reduces the amount of
contrast reducing saturation of the adjacent slices.
"sinc"
profile
computer -optimized
profile
"rectangular"
profile
overlapped area
1
Interslice gap
Interleaving
10
Bad pixel in k-space
479
Resultant image
FIGURE 13-28 A single bad pixel in k-space causes a significant artifact in the reconstructed image. The
bad pixel is located at kx 2, ky 3, which produces a superimposed sinusoidal wave on the spatial domain
image as shown above.
Motion Artifacts
PEG
FEG
The most ubiquitous and noticeable artifacts in MRI arise with patient motion, including
voluntary and involuntary movement, and flow (blood, CSF). Although motion artifacts
are not unique to MRI, the long acquisition time of certain MRI sequences increases the
probability of motion blurring and contrast resolution losses. Motion artifacts occur
mostly along the phase encode direction, as adjacent phase encoding measurements in
k-space are separated by a TR interval that can last 3,000 ms or longer. Even very slight
motion can cause a change in the recorded phase variation across the FOV throughout
the MR acquisition sequence. Examples of motion artifacts are shown in Figure 13-29.
The frequency encode direction is less affected, especially by periodic motion, since the
evolution of the echo signal, frequency encoding, and sampling occur simultaneously
over several milliseconds. Ghost images, which are faint copies of the image displaced
along the phase encode direction, are the visual result of patient motion.
Several techniques can compensate for motion-related artifacts. The simplest
technique transposes the PEG and FEG to relocate the motion artifacts out of the
PEG
FEG
PEG
FEG
PEG
PEG
FEG
FEG
FIGURE 13-29 Motion artifacts, particularly of flow patterns, are most always displayed in the phase encode
gradient direction. Slight changes in phase produce multiple ghost images of the anatomy, since the variation
in phase caused by motion can be substantial between excitations.
480
region of diagnostic interest with the same pulse sequences. This does not reduce the
magnitude of the artifacts, however, and often there is a mismatch when placing the
PEG along the long axis of a rectangular FOV (e.g., an exam of the thoracic spine)
in terms of longer examination times or a significant loss of spatial resolution or of
SNR.
There are other motion compensation methods:
1. Cardiac and respiratory gatingsignal acquisition at a particular cyclic location
synchronizes the phase changes applied across the anatomy (Fig. 13-30).
2. Respiratory ordering of the phase encoding projections based on location within the
respiratory cycle. Mechanical or video devices provide signals to monitor the cycle.
3. Signal averaging to reduce artifacts of random motion by making displaced signals less conspicuous relative to stationary anatomy.
4. Short TE spin echo sequences (limited to proton density, T1-weighted scans, fractional echo acquisition, Fig. 13-3). Note: Long TE scans (T2 weighting) are more
susceptible to motion.
5. Gradient moment nulling (additional gradient pulses for flow compensation) to
help rephase protons that are dephased due to motion. Most often, these techniques
require a longer TE and are more useful for T2-weighted scans (Fig. 13-18).
6. Presaturation pulses applied outside the imaging region to reduce flow artifacts
from inflowing protons, as well as other patient motions that occur in the periphery (Fig. 13-12).
7. Multiple redundant sampling in the center of k-space (e.g., propeller) to identify
and remove those sequences contributing to motion, without deleteriously affecting the image (Fig. 13-8).
Chemical Shift Artifacts of the First Kind

There are two types of chemical shift artifacts that affect the display of anatomy due
to the precessional frequency differences of protons in fat versus protons in water.
Chemical shift refers to the resonance frequency variations resulting from intrinsic
Non Gated
ECG Signal
Velocity
5 6
1 2 3
2 3 4
1 2 3
4
4
3
5 6 1 2
ECG Gated
ECG Signal
Velocity
2
1
2 3
5 6
FIGURE 13-30 Motion artifacts occur when data is acquired without consideration of physiologic periodicity. Top. The electrocardiogram measures the R-wave at each heartbeat, but data acquisition proceeds in a
linear fashion without regard to reproducibility. The result is a set of images degraded with motion artifact,
with diagnostic usefulness marginal, at best. Bottom. Acquisition of images proceeds with the detection of
the R-wave signal and synchronization of the collection of image data in a stepwise fashion over the period
between R-waves. A reduced number of images or extended acquisition time is required to collect the data.
481
magnetic shielding of anatomic structures. Molecular structure and electron orbital

characteristics produce fields that shield the main magnetic field and give rise to distinct peaks in the MR spectrum. In the case of proton spectra, peaks correspond to
water and fat, and in the case of breast imaging, silicone material is another material
to consider. Lower frequencies of about 3.5 parts per million for protons in fat and
5.0 parts per million for protons in silicone occur, compared to the resonance frequency
of protons in water (Fig. 13-31). Since resonance frequency increases linearly with field
strength, the absolute difference between the fat and water resonance also increases,
making high field strength magnets more susceptible to chemical shift artifact.
Data acquisition methods cannot directly discriminate a frequency shift due to
the application of a frequency encode gradient or a chemical shift artifact. Water and
fat differences therefore cannot be distinguished by the frequency difference induced
by the gradient. The protons in fat resonate at a slightly lower frequency than the
corresponding protons in water, and cause a shift in the anatomy (misregistration of
water and fat moieties) along the frequency encode gradient direction.
A sample calculation in the example below demonstrates frequency variations in
fat and water for two different magnetic field and gradient field strengths:
Chemical shift artifact numerical calculation for field strength, with a 3.5-ppm
(3.5 106) variation in resonance frequency between fat and water results in the
following frequency differences:
1.5 T: 63.8 106 Hz 3.5 10 6 223 Hz
3.0 T:127.7 106 Hz 3.5 10 6 447 Hz
Thus, the chemical shift is more severe for higher field strength magnets.
Chemical shift artifact numerical calculation for gradient strength results in the
following numerical calculations for a 25-cm (0.25 m) FOV, 256 256 matrix:
Low gradient strength: 2.5 mT/m 0.25 m 0.000625 T variation, gives frequency
range of 0.000625 T 42.58 MHz/T 26.6 kHz across FOV and 26.6 kHz/256
pixels 104 Hz/pixel
High gradient strength: 10 mT/m 0.25 m 0.0025 T variation, gives frequency
range of 0.0025 T 42.58 MHz/T 106.5 kHz across FOV and 106.5 kHz/256
pixels 416 Hz/pixel.
Thus, a chemical shift occurrence is more severe for lower gradient strengths, since
displacement will occur over a large number of pixels. With a higher gradient
strength, water and fat are more closely contained within the broader pixel boundary
bandwidths. Normal and low bandwidth images are illustrated in Figure 13-32.
silicone
fat
Chemical Shift
water
3-4 ppm difference, fat

~5 ppm difference, silicone
frequency
+
+
water
water
FEG
PEG
fat
fat
-
PEG
+
FEG
FIGURE 13-31 Chemical shift refers to

the slightly different precessional frequencies of protons in different materials or tissues. The shifts (in ppm) are referenced to
water for fat and silicone. Fat chemical shift
artifacts are represented by a shift of water
and fat in the images of anatomical structure, mainly in the frequency encode gradient direction. Swapping the PEG and the
FEG will cause a shift of the fat and water
components of the tissues in the image.
482
High Bandwidth
Low Bandwidth
PEG
FEG
FIGURE 13-32 MR images of the breast, containing glandular and adipose tissue, are acquired under a high
bandwidth (32 kHz) and a low bandwidth (4 kHz), illustrating the more severe chemical shift with low readout
gradient strength and bandwidth. (Reprinted by permission, Hendrick RE. Breast MRI: fundamentals and technical aspects. New York, NY: Springer, 2007.)
RF bandwidth and gradient strength considerations can mitigate chemical shift artifacts. While higher gradient strength can confine the chemical shift of fat within the pixel
bandwidth boundaries, a significant SNR penalty occurs with the broad RF bandwidth
required to achieve a given slice thickness. A more widely used method is to use lower
gradient strengths and narrow bandwidths in combination with off-resonance chemical presaturation RF pulses to minimize the fat (or the silicone) signal in the image (Fig.
13-33). Another alternative is to use STIR techniques to eliminate the signals due to fat
at the bounce point. Swapping the phase and frequency gradient directions or changing the polarity of the frequency encode gradient can displace chemical shift artifacts
from a specific image region, even though the chemical shift displacement still exists.
Identification of fat in a specific anatomic region is easily discerned from the chemical
shift artifact displacement caused by changes in FEG/PEG gradient directions.
Chemical Shift Artifacts of the Second Kind

Chemical shift artifacts of the second kind occur with GE images, resulting from the rephasing and dephasing of the echo in the same direction relative to the main magnetic field.
Signal appearance is dependent on the selection of TE. This happens because of constructive (in phase) or destructive (out of phase) transverse magnetization events that occur
periodically due to the difference in precessional frequencies. At 1.5 T, the chemical shift
FIGURE 13-33 The left image of the lumbar
spine is Spin Echo T1 weighted, TR 450 ms,
TE 14 ms. The right image is T1 weighted
with chemical fat saturation pulses, TR 667
ms, TE 8 ms. In both images, the FEG is vertically oriented.
FEG
Water
Fat
483
Water & Fat In-Phase Every 4.4 ms
2.5
2
TE=
7 ms
1.5
1
Fat Water
TE =
9 ms
0.5
0
0
10
15
TE (ms)
Water & Fat Out-of-Phase at 2.2, 6.6, 11.0 ms,

FIGURE 13-34 For GE image sequences, signal intensity of the transverse magnetization vector due to
the 220 Hz lower precessional frequency of fat protons, where in-phase magnetization occurs every 4.4 ms
(1/220s1), and out-of-phase magnetization occurs every 4.4 ms shifted by cycle (2.2 ms). Signal intensity is
dependent on the selection of TE, as shown above.
is 220 Hz, and the periodicity of each peak (in phase) between water and fat occurs at 0,
4.5, 9.0, 13.5, . ms, and each valley (out of phase) at 2.25, 6.75, 11.0, . ms, as shown
in Figure 13-34. Thus, selection of TE at 9 ms will lead to a constructive addition of water
and fat, and TE at 7 ms will lead to a destructive addition of water and fat. The in-phase
timing will lead to a conventional chemical shift image of the first kind, while the out-ofphase timing will lead to a chemical shift image of the second kind, manifesting a dark rim
around heterogeneous water and fat anatomical structures, shown in Figure 13-35.
Ringing Artifacts
Ringing artifact (also known as Gibbs phenomenon) occurs near sharp boundaries
and high-contrast transitions in the image, and appears as multiple, regularly spaced
parallel bands of alternating bright and dark signal that slowly fades with distance.
TE = 9 ms
Water & Fat in-phase
TE = 7 ms
Water & Fat out-of-phase
PEG
FEG
FIGURE 13-35 Breast MRI images show the effect of selecting a specific TE for a GE acquisition. On the left,
chemical shift of the first kind is shown with TE 9 ms and water and fat in phase for transverse magnetization, shifted only due to the intrinsic chemical shift differences of fat and water. On the right, chemical shift
of the second kind is additionally manifested with TE 7 ms, due to fat and water being out of phase, creating
a lower signal at all fat-water interfaces, and resulting in reduced intensity. (Reprinted by permission, Hendrick
RE. Breast MRI: fundamentals and technical aspects. New York, NY: Springer, 2007.)
484
The cause is the insufficient sampling of high frequencies inherent at sharp discontinuities in the signal. Images of objects can be reconstructed from a summation of sinusoidal waveforms of specific amplitudes and frequencies, as shown in Figure 13-36 for
a simple rectangular object. In the figure, the summation of frequency harmonics, each
with a particular amplitude and phase, approximates the distribution of the object, but
initially does very poorly, particularly at the sharp edges. As the number of higher frequency harmonics increase, a better estimate is achieved, although an infinite number
of frequencies are theoretically necessary to reconstruct the sharp edge perfectly.
In the MR acquisition, the number of frequency samples is determined by the
number of pixels (frequency, kx, or phase, ky, increments) across the k-space matrix.
For 256 pixels, 128 discrete frequencies are depicted, and for 128 pixels, 64 discrete
frequencies are specified (the k-space matrix is symmetric in quadrants and duplicated about its center). A lack of high-frequency signals causes the ringing at sharp
transitions described as a diminishing hyper- and hypointense signal oscillation from
the transition. Ringing artifacts are thus more likely for smaller digital matrix sizes
(Fig. 13-37, 256 versus 128 matrix). Ringing artifact commonly occurs at skull/brain
interfaces, where there is a large transition in signal amplitude.
A.
Frequency
harmonics
1st
2
1
0
-1
-2
1st + 3rd
1st + 3rd + 5th
1st + 3rd + 5th + 7th
B.
Spatial
domain
rectangle
estimate
2
1
0
-1
-2
-1
-1
-2
-2
-1
-1
-2
-2
-1
-1
-2
-2
FIGURE 13-36 The synthesis of a spa- C. Sharp transition in MR image:

tial object occurs by the summation of
frequency harmonics in the MR image.
A.Left column: frequency harmonics that
estimate a rectangle function with progressively higher frequencies and lower ampli- Ringing
tudes are shown. B. Middle column: As
n = 128
higher frequency harmonics are included,
samples
the summed result more faithfully represents the object shape, in this example
a rectangle with two vertical edges. The
number of frequencies encoded in the MR
image is dependent on the matrix size.
C.Right column: A sharp transition boundary in an MR image is represented with 256
n=256
samples
samples better than with128 samples (frequency harmonics in k-space). The amount
of ringing caused by insufficient sampling is
reduced with a larger number of samples.
256 (vertical) 128 (horizontal)
256 256
ringing
485
FIGURE 13-37 Example of ringing

artifacts caused by a sharp signal transition at the skull in a brain image for
a 256 128 matrix (left) along the
short (horizontal) axis, and the elimination of the artifact in a 256 256
matrix (right). The short axis defines
the PEG direction.
Wraparound Artifacts
The wraparound artifact is a result of the mismapping of anatomy that lies outside of the FOV but within the slice volume. The anatomy is usually displaced to
the opposite side of the image. It is caused by nonlinear gradients or by undersampling of the frequencies contained within the returned signal envelope. For
the latter, the sampling rate must be twice the maximal frequency that occurs in
the object (the Nyquist sampling limit). Otherwise, the Fourier transform cannot distinguish frequencies that are present in the data above the Nyquist frequency limit, and instead assigns a lower frequency value to them (Fig. 13-38).
Frequency signals will wraparound to the opposite side of the image, masquerading as low-frequency (aliased) signals.
In the frequency encode direction, a low-pass filter can be applied to the acquired
time domain signal to eliminate frequencies beyond the Nyquist frequency. In the
phase encode direction, aliasing artifacts can be reduced by increasing the number
of phase encode steps (the trade-off is increased image time). Another approach is to
move the region of anatomic interest to the center of the imaging volume to avoid the
FOV
C
Wrap-around
Sampling rate
FIGURE 13-38 Left. Wraparound artifacts are caused by aliasing. Shown is a fixed sampling rate and net precessional frequencies occurring at position A and position B within the FOV that have identical frequencies but different phase. If signal from position C is at twice the frequency of B and insufficiently sampled, the same frequency
and phase will be assigned to C as that assigned to A, and therefore will appear at that location. Right. A wraparound artifact example displaces anatomy from one side of the image (or outside of the FOV) to the other side.
486
overlapping anatomy, which usually occurs at the periphery of the FOV. An antialiasing saturation pulse just outside of the FOV is yet another method of eliminating
high-frequency signals that would otherwise be aliased into the lower frequency spectrum. This example of wrap-around artifact is easy to interpret. In some cases, the
artifact is not as well delineated (e.g., the top of the skull wrapping into the brain).
Partial Volume Artifacts

Partial volume artifacts arise from the finite size of the voxel over which the signal is
averaged. This results in a loss of detail and spatial resolution. Reduction of partial
volume artifacts is accomplished by using a smaller pixel size and/or a smaller slice
thickness. With a smaller voxel, the SNR is reduced for a similar imaging time, resulting in a noisier signal with less low-contrast sensitivity. Of course, with a greater NEX
(averages), the SNR can be maintained, at the cost of longer imaging time.
13.6 Magnetic Resonance Spectroscopy

Magnetic resonance spectroscopy (MRS) is a method to measure tissue chemistry (an
electronic biopsy) by recording and evaluating signals from metabolites by identifying metabolic peaks caused by frequency shifts (in parts per million, ppm) relative to a frequency standard. In vivo MRS can be performed with 1H (proton), 23Na
(sodium), and 31P (phosphorus) nuclei, but proton spectroscopy provides a much
higher SNR and can be included in a conventional MRI protocol with about 10 to 15
min extra exam time. Uses of MRS include serial evaluation of biochemical changes
in tumors, analyzing metabolic disorders, infections and diseases, as well as evaluation of therapeutic oncology treatments for tumor recurrence versus radiation damage. Early applications were dedicated to brain disorders, but now breast, liver, and
prostate MRS are also performed. Correlation of spectroscopy results and MR images
are always advised before making a final diagnosis.
In MRS, signals are derived from the amplitude of proton metabolites in targeted
tissues. In these metabolites, chemical shifts occur due to electron cloud shielding
of the nuclei, causing slightly different resonance frequencies, which exist in a frequency range between water and fat. The very small signal amplitudes of the metabolites require suppression of the extremely large (10,000 times higher) amplitudes
due to bulk water and fat protons, as shown in Figure 13-39. This is achieved by
using specific chemical saturation techniques, such as CHESS (Chemical Shift-Selective) or STIR (see Chapter 12). In many cases, the areas evaluated are away from fat
structures, and only bulk water signal suppression is necessary; however, in organs
such as the liver and the breast, suppression of both fat and water are required. Once
the water and fat signals are suppressed, localization of the targeted area volume is
achieved by either a single voxel or multivoxel technique.
Single voxel MRS sampling areas, covering a volume of about 1 cm3 are delineated by a STEAM (Stimulated echo acquisition mode) or a PRESS (Point Resolved
Spectroscopy) sequence. The STEAM method uses a 90-degree excitation pulse and
90-degree refocusing pulse to collect the signal in conjunction with gradients to
define each dimension of the voxel. The PRESS sequence uses a 90-degree excitation
and 180-degree refocusing pulse in each direction. STEAM achieves shorter echo
times and superior voxel boundaries, but with lower SNR. After the voxel data are collected, a Fourier transform is applied to separate the composite signal into individual
frequencies, which are plotted as a trace for a normal brain spectrum (Fig.13-40).
The resulting line widths are based on homogeneity of the main magnetic field as
FIGURE 13-39 MRS metabolites of

interest in comparison to the water and
fat peaks commonly used for imaging.
In order to isolate the very small signals,
chemical saturation of the water (and fat
when present) signal is essential.
Water peak
Relative Amplitude
487
Metabolites
of interest
Fat peak
Frequency shift (ppm)
RadioGraphics 2006; 26:S173S189
well as the magnetic field strength. Higher field strengths (e.g., 3.0 T) will improve
resolution of the peaks and corresponding SNR.
Multivoxel MRS uses a CSI (Chemical Shift Imaging) technique to delineate multiple
voxels of approximately 1 cm3 volume in 1, 2, or 3 planes over a rectangular block of several centimeters, achievable with more sophisticated equipment and longer scan times.
This is followed by MRSI (Magnetic Resonance Spectroscopic Imaging) where the signal
intensity of a single metabolite in each voxel is color encoded for each voxel according
to concentration and the generated parameter maps superimposed on the anatomical
MR image. In practice, the single voxel technique is used to make the initial diagnosis
because the SNR is high and all metabolites are represented in the MRS trace. Then, a
multivoxel acquisition to assess the distribution of a specific metabolite is performed.
Proton MRS can be performed with short (20 to 40 ms), intermediate (135 to
145 ms), or long (270 to 290 ms) echo times. For short TE, numerous resonances
from metabolites of lesser importance (with shorter T2) can make the spectra more
difficult to interpret, and with long echo times, SNR losses are too severe. Therefore,
most MRS acquisitions use a TE of approximately 135 ms at 1.5 T. Metabolites of
interest for brain spectroscopy are listed in Table 13-2.
Applications of MRS are achieved through the interpretation of the spectra that
are obtained from the lesion, from its surroundings, and presumably healthy tissue in
MR Spectrum from anaplastic oligoastrocytoma Choline / Creatine ratio map

FIGURE 13-40 Left: Intermediate echo (TE=135 ms) single voxel spectrum is shown, positioned over an anaplastic
oligoastrocytoma brain lesion. Note the elevated Choline peak and lowered Creatine and NAA peaks. Right: Multivoxel spectrum is color coded to the Choline / Creatine ratio, illustrating the regional variation of the metabolites corresponding to tumor. From Al-Okaili RN, Krejza J, Wang S, Woo JH, Melhem ER. Advanced MR Imaging Techniques
in the Diagnosis of Intraaxial Brain Tumors in Adults. Radiographics 2006; 26: S173-S189. By permission.
488
TABLE 13-2 METABOLITES IN MRS AT 1.5 T

PROPERTIES/SIGNIFICANCE
IN THE BRAIN
ABBREVIATION
METABOLITE
SHIFT (PPM)
Cho
Phosphocholine
3.22
Membrane turnover, cell

proliferation
Cr
Creatine
3.02 and 3.93
Temporary store for

energy-rich phosphates
NAA
N-acetyl-l-aspartate
2.01
Presence of intact glioneural structures
1.33 (inverted)
Anaerobic glycolysis
1.21.4
Necrosis
Lactate
Lipids
Free fatty acids
the same scan. Main criteria include the presence or absence of pathologic metabolites (lactate or lipids) and the relationships between the concentrations of choline,
creatine, and NAA as ratios. Spectra are usually scaled to the highest peak, and y-axis
values will usually differ between measurements, so caution must be observed to
avoid potential misdiagnoses.
In a tumor, high cell turnover causes an increase in choline concentration along
with a corresponding depression of the NAA peak caused by the loss of healthy
glioneural structures. In addition, the creatine peak may also be reduced depending
on the energy status of the tumor; it is often used to serve as an internal reference
for calculating ratios of metabolites. When a lipid peak is observed, this is a sign of
hypoxia and the likelihood of a high-grade malignancy. Table 13-3 lists qualitative
findings for MRS spectroscopy in evaluating brain tumors for the ratios of NAA/Cr,
NAA/Cho, and Cho/Cr. Examples of single voxel and multivoxel spectra are shown
in Figure 13-40, illustrating the use of the spectral peaks for diagnostic interpretation
of the ratios, and a MRSI color-encoded graphic display of the spatial distribution
of findings. There is certainly much more than determining ratios, and differential
diagnoses are often clouded by indistinct peaks, poor SNR, and tumor heterogeneity,
among other causes. MRS at 3 T field strength enjoys much better SNR, improved
spectral resolution, and faster scans. Because of the higher spectral resolution, familiar single target peaks at 1.5 T become a collection of peaks at 3 T.
13.7 Ancillary Components

Ancillary components are necessary for the proper functioning of the MR system.
Shim coils are active or passive magnetic field devices that are used to adjust the main
magnetic field and to improve the homogeneity in the sensitive central volume of the
TABLE 13-3RATIOS OF METABOLITE PEAKS IN
MRS INDICATING NORMAL AND

ABNORMAL STATUS
METABOLITE RATIO
NORMAL
ABNORMAL
NAA/Cr
2.0
1.6
NAA/Cho
1.6
1.2
Cho/Cr
1.2
.1.5
489
scanner. Gradient coils, as previously discussed, are wire conductors that produce a
linear superimposed gradient magnetic field on the main magnetic field. The gradient
coils are located inside the cylinder of the magnet, and are responsible for the banging
noise one hears during imaging. This noise is caused by the flexing and torque experienced by the gradient coil from the rapidly changing magnetic fields when energized.
Radiofrequency Coils
A wide variety of transmit and receive coils complement a MR scanner. RF transmitter coils create an oscillating secondary magnetic field formed by passing an alternating current through a loop of wire. To accomplish excitation and resonance, the
created secondary B1 field must be arranged at right angles to the main magnetic field,
B0. In an air core design with a horizontal field, the RF coil secondary field should
be in the transverse or vertical axes, as the B1 field is created perpendicular to the
transmit coils themselves. RF transmitter coils are therefore oriented above, below, or
at the sides of the patient, and are usually cylindrical. In most systems, the body coil
contained within the bore of the magnet is most frequently used, but also transmitter
coils for the head, extremity, and some breast coils are coupled to a receiver coil.
Very often, transmit and receive functions are separated to handle the variety of
imaging situations that arise, and to maximize the SNR for an imaging sequence.
All RF receiver coils must resonate and efficiently store energy at the Larmor frequency. This is determined by the inductance and capacitance properties of the coil.
RF transmit and receive coils need to be tuned prior to each acquisition and matched
to accommodate the different magnetic inductance of each patient. Receiver coils
must be properly placed to adequately detect the MR signal.
Proximity RF coils include volume or bird-cage coils, the design of choice for brain
imaging, the single-turn solenoid for imaging the extremities and the breasts, and the
saddle coil. These coils are typically operated as both a transmitter and receiver of RF
energy (Fig. 13-41). Volume coils encompass the total area of the anatomy of interest and yield uniform excitation and SNR over the entire imaging volume. However,
because of their relatively large size, images are produced with lower SNR than other
types of coils. Enhanced performance is obtained with a process known as quadrature
excitation and detection, which enables the energy to be transmitted and the signals to
be received by two pairs of coils oriented at right angles, either electronically or physically. This detector manages two simultaneous channels known as the real (records
MR information in phase with a reference signal) and the imaginary (records MR information 90 degree out of phase with the reference signal) channels, and increases the
SNR up to a factor of 2 . If imbalances in the offset or gain of these detectors occur,
then artifacts will be manifested, such as a center point artifact.
Surface coils are used to achieve high SNR and high resolution when imaging
anatomy near the surface of the patient, such as the spine. They are typically receiveonly designs, and are usually small and shaped for a specific imaging exam and
for patient comfort. The received signal sensitivity, however, is limited to the volume located around the coil at a depth into the patient equal to the radius of the
coil, which causes a loss of signal with depth. There are now intracavitary coils for
endorectal, endovascular, endovaginal, esophageal, and urethral local imaging, and
can be used to receive signals from deep within the patient. In general, a body coil is
used to transmit the RF energy and the local coil is used to receive the MR signal.
Phased array coils consisting of multiple coils and receivers are made of several
overlapping loops, which extend the imaging FOV in one direction (Fig. 13-41). The
490
FIGURE 13-41 Radiofrequency surface coils improve image quality and SNR for specific examinations. Upper
left is a dedicated head and neck coil. Upper right is a dedicated head coil; the lower left and lower right images
illustrate the two components of an 8 channel phased array coil for body imaging.
small FOV of each individual coil provides excellent SNR and resolution, and each
is combined to produce a composite image with the advantages of the local surface
coil, so that all data can be acquired in a single sequence. Phased array coils for the
spine, pelvis, breast, cardiac, and temperomandibular joint applications are commonly purchased with an MR system for optimal image quality.
Multi-channel encoding coils with as many as N 32 elements allow for detection
and encoding based upon the detection of a sensitivity map of the signals near the coil.
These coils are used in parallel imaging to fill N multiple lines of k-space per TR interval,
by assigning certain coil responses to specific regions as the data are acquired simultaneously and using software to link them electronically. By filling k-space quicker, the
scan time can be reduced by the number of elements in the coil. However, since each
line of k-space is encoded by separate coils, the gap between each line for a specific coil
is N times greater than if k-space had been filled normally. Since the FOV dimension
in the phase encode direction is inversely proportional to the spacing, the size of the
FOV is reduced to 1/N its original size, and as a result, aliasing of signals outside of the
FOV in the phase encode direction occurs, and each coil response produces a wrapped
image. To overcome the aliasing, the system uses the sensitivity profile of each coil to
calculate where the signal is coming from relative to the coil based on its amplitude
the signal generated near the coil has a higher amplitude than the signal furthest away.
Using this process (called SENSE, ASSET, GRAPPA, iPAT by the various manufacturers), the image is unwrapped and combined with the unwrapped images from the
other coils to form one summed image of the slice, with high SNR and resolution
(see Figure 13-9). Parallel imaging can be used to either reduce scan time or improve
resolution, in conjunction with most pulse sequences. Downsides include image misalignment when combining images due to patient motion, and increase in chemical
shift artifacts, due to a range of frequencies mapped across each coil. Nevertheless,
multi-coil parallel imaging techniques are now common and increasing in use.
491
RF coil safety requires regular inspection of the coil condition, including the conductive wires leading to the coils from the connectors. These wires have the capacity
to transmit heat, which may burn the insulating material of the wires or burn the
patient. It is important to ensure that the coils are not looped and do not touch the
patient or the bore of the magnet. Damage to the insulation requires immediate repair,
as this could result in extreme heating, fire, and potential harm to the patient.
MR System Subcomponents
The control interfaces, RF source, detector, and amplifier, analog to digital converter
(digitizer), pulse programmer, computer system, gradient power supplies, and image
display are crucial components of the MR system. They integrate and synchronize the
tasks necessary to produce the MR image (Fig. 13-42).
The operator interface and computer systems vary with the manufacturer, but
most consist of a computer system, dedicated processor for Fourier transformation,
image processor to form the image, disk drives for storage of raw data and pulse
sequence parameters, and a power distribution system to distribute and filter the
direct and alternating current. The operators console is located outside of the scan
room, and provides the interface to the hardware and software for data acquisition.
13.8 Magnet Siting, Quality Control
Magnet Siting
Superconductive magnets produce extensive magnetic fringe fields, and create potentially hazardous conditions in adjacent areas. In addition, extremely small signal
amplitudes generated by the protons in the body during an imaging procedure have
a frequency common to commercial FM broadcasts. Thus, two requirements must
MRI system components

Shim Power
Supply
Magnet
RF Transmitter
and Receiver
Patient
Table
Data
Storage
Digitizer &
Image
Processor
Clock
Gradient
Pulse Prog
Gradient
Power Supply
Pulse Prog
Measurement
Control
Host
Computer
Operating
Console
FIGURE 13-42 A block diagram of a typical MRI system shows the components and interfaces between
subsystems. Not shown is the Faraday cage that surrounds the magnet assembly to eliminate environmental RF noise.
492
be considered for MR system siting: protect the local environment from the magnet
system and protect the magnet system from the local environment.
Fringe fields from a high field strength magnet can extend quite farroughly equal
to B0, where is a constant dependent on the magnet bore size and magnet configuration. Fringe fields can potentially cause a disruption of electronic signals and sensitive
electronic devices. An unshielded 1.5-T magnet has a 1-mT fringe field at a distance of
approximately 9.3 m, a 0.5-mT field at 11.5 m, and a 0.3-mT field at 14.1 m from the
center of the magnet. Magnetic shielding is one way to reduce fringe field interactions in
adjacent areas. Passive (e.g., thick metal walls close to the magnet) and active (e.g., electromagnet systems strategically placed in the magnet housing) magnetic shielding systems permit a significant reduction in the extent of the fringe fields for high field strength,
air core magnets (Fig. 13-43). Patients with pacemakers or ferromagnetic aneurysm clips
must avoid fringe fields above 0.5 mT. Magnetically sensitive equipment such as image
intensifiers, gamma cameras, and color TVs are severely impacted by fringe fields of less
than 0.3 mT, as electromagnetic focusing in these devices is disrupted.
Administrative control for magnetic fringe fields is 0.5 mT, requiring controlled
access to areas that exceed this level. Magnetic fields below 0.5 mT are considered
safe for the patient population. Disruption of the fringe fields can reduce the homogeneity of the active imaging volume. Any large metallic object (elevator, automobile,
etc.) traveling through the fringe field can produce such an effect.
Environmental RF noise must be reduced to protect the extremely sensitive receiver
within the magnet from interfering signals. The typical approach for stray RF signal
protection is the construction of a Faraday cage, an internal enclosure consisting of RF
attenuating copper sheet and/or copper wire mesh. The room containing the MRI system is typically lined with copper sheet (walls) and mesh (windows). This is a costly
construction item but provides effective protection from stray RF noise (Fig. 13-44).
Field Uniformity
Radial distance (m)
In addition to magnetic field strength, field uniformity is an important characteristic,

expressed in parts per million (ppm) over a given volume, such as 40 cm3. This is
based upon the precessional frequency of the proton, which is determined from the
12
0.3 mT
10
0.5
3
5
Unshielded
20
2
2
2
4
1
0.5
0.3
10 12 14
Axial distance (m)
Shielded
0.1 mT
8
FIGURE 13-43 An unshielded (top half of diagram) and shielded (bottom half of diagram) 1.5 T magnet and
the magnetic fringe field strengths plotted with radial distance (vertical axis) and axial distance (horizontal axis).
493
FIGURE 13-44 The MR scanner room requires protection from extraneous radiofrequency signals. This is
achieved with the installation of a Faraday cage comprised of copper sheet that lines the inner walls of
the room (left), copper mesh covering the operator viewing window (not shown), and a copper lined door
and doorjamb with an inflatable bladder conductor (note switch above the door handle) to seal the door
(middle and upper right). A leak in the Faraday cage will result in RF artifacts that will occur at specific
frequencies as streaks across the image, perpendicular to the FEG direction. (Lower right. FEG is vertical,
PEG is horizontal.)
proton gyromagnetic ratio. At 1.5 T, the precessional frequency is 1.5 T 42.58

MHz/T 63.8 MHz 63.8 106 cycles/s, and a specification of 2 ppm homogeneity
(2 10 6 ) gives a frequency uniformity of about 128 cycles/s (Hz) over the volume.
Typical homogeneities range from less than 1 ppm for a small FOV (e.g. 150 mm)
to greater than 10 ppm for a large FOV (e.g., 400 mm). Field uniformity is achieved
by manipulating the main field peripherally with passive and active shim coils,
which exist in proximity to the main magnetic field. These coils interact with the
fringe fields and adjust the variation of the central magnetic field.
Quality Control
Like any imaging system, the MRI scanner is only as good as the weakest link in the
imaging chain. The components of the MR system that must be periodically checked
include the magnetic field strength, magnetic field homogeneity, system field shimming, gradient linearity, system RF tuning, receiver coil optimization, environmental
noise sources, power supplies, peripheral equipment, and control systems among
others. A QC program should be designed to assess the basic day-to-day functionality of the scanner. A set of QC tests is listed in Table 13-4. Qualitative and quantitative measurements of system performance should be obtained on a periodic basis,
with a test frequency dependent on the likelihood of detecting a change in the baseline values outside of normal operating limits.
The American College of Radiology has an MRI accreditation program that specifies
requirements for system operation, QC, and the training requirements of technologists, radiologists, and physicists involved in scanner operation. The accreditation process evaluates the qualifications of personnel, equipment performance, effectiveness
of QC procedures, and quality of clinical imagesfactors that are consistent with the
maintenance of a state-of-the-art facility.
494
TABLE 13-4RECOMMENDED QC TESTS FOR MRI
SYSTEMS
High-contrast spatial resolution
Slice thickness accuracy
Slice position accuracy
RF center frequency tuning
Geometric accuracy and spatial uniformity
Signal uniformity
Low-contrast resolution (sensitivity)
Image artifact evaluation
Operational controls (e.g., table movement control and alignment
lighting checks)
Preventive maintenance logging and documentation
Review of system log book and operations
MRI phantoms are composed of materials that produce MR signals with carefully
defined relaxation times. Some materials are aqueous paramagnetic solutions; pure
gelatin, agar, silicone, or agarose; and organic doped gels, paramagnetic doped gels,
and others. Water is most frequently used, but it is necessary to adjust the T1 and T2
relaxation times of (doped) water, so that images can be acquired using pulse sequence
timing for patients (e.g., this is achieved by adding nickel, aqueous oxygen, aqueous manganese, or gadolinium). For example, the ACR MR Accreditation phantom
(Fig.13-45) is a cylindrical phantom of 190 mm inside diameter and 148 mm inside
length. It is filled with a 10 millimolar solution of nickel chloride and 75 millimolar
sodium chloride. Inside the phantom are several structures that are used in a variety of
tests for scanner performance. In this phantom, seven quantitative tests are made by
scanning the phantom with specific instructions, and include geometric accuracy, high
contrast spatial resolution, slice thickness accuracy, slice position accuracy, image intensity uniformity, percent signal ghosting, and low-contrast object detectability. Details
MRI Accreditation
Phantom
FIGURE 13-45 The ACR accreditation phantom (upper left) and selected images scanned from the phantom are shown above. Upper middle is the spatial resolution module, and just below is a magnification image
of the test targets. The upper right shows the geometric distortion module, the lower right is the low contrast
sensitivity module, and the lower left is the uniformity module. Not all images are shown; specific details are
described in the accreditation documentation available on the ACR website, http://www.acr.org.
495
on measurement analysis, recommended action criteria, and causes of failure/corrective action are included in the guidance. Some phantoms have standards with known
T1, T2, and proton density values to evaluate the quantitative accuracy of the scanner,
and to determine the ability to achieve an expected contrast level for a given pulse
sequence. Homogeneity phantoms determine the spatial uniformity of transmit and
receive RF magnetic fields. Ideal performance is a spatially uniform excitation of the
protons and a spatially uniform sensitivity across the imaged object.
13.9 MR Bioeffects and Safety

Although ionizing radiation is not used with MRI and MRS, perception of the general
public is that MR is a very safe modality, safety aspects are often an afterthought,
particularly in terms of operational activities and training of personnel who work
in the area and around the magnet. There are very many important bioeffects and
safety issues to be considered for MR. These include the presence of strong magnetic
fields, RF energy, time-varying magnetic gradient fields, cryogenic liquids, a confined imaging device (claustrophobia), and noisy operation (gradient coil activation
and deactivation, creating acoustic noise). Patients with implants, prostheses, aneurysm clips, pacemakers, heart valves, etc., should be aware of considerable torque on
the devices which when placed in the magnetic field, could cause serious adverse
effects. Nonmetallic implant materials can also lead to significant heating under rapidly changing gradient fields. Consideration of the distortions and artifacts on the
acquired images and possibility of misdiagnosis are also a concern. Ferromagnetic
materials inadvertently brought into the imaging room (e.g., an IV pole) are attracted
to the magnetic field and can become a deadly projectile to the occupant within the
bore of the magnet. Many unfortunate deaths have been attributed to carelessness
and lack of a safety culture around an MR scanner. Signage exists in three categories
to help identify MR-compatible materials. MR safe is a square green sign, and is put
on materials and objects that are wholly nonmetallic; not MR safe is round red, and
is placed on all ferromagnetic and many conducting metals; MR conditional signage is triangular yellow, placed on objects that may or may not be safe until further
investigation is performed. These signs are shown in Figure 13-46.
In extreme emergencies, the superconducting magnet can be turned off by a
manually controlled quench procedure. Even under the best circumstances, the
quench procedure subjects the magnet to a 260 degrees temperature difference in a
short period of time. If performed too quickly, major physical damage to the magnet
FIGURE 13-46 MR labeling includes on the left, MR safe materials that have been found not to interact
with the strong MR field or disrupt operation; in the middle, NOT MR safe labels that contraindicate bringing
materials with this designation past Zone 2 (see discussion later in this section; on the right, more consideration required before bringing such labeled objects into the scan room).
496
TABLE 13-5 MRI SAFETY GUIDELINES

ISSUE
PARAMETER
VARIABLES
SPECIFIED VALUE
Static magnetic field
Magnetic field (B0)
Maximum strength
3.0 T
Inadvertent exposure
Maximum
0.0005 T
Changing magnetic
field (dB/dt)
Axial gradients
.120 s
,20 T/s
12 s , ,120 s
,2400/ (s) T/s
,12 s
,200 T/s
RF power deposition
Transverse gradients
,3 axial
gradients
System
,6 T/s
Temperature
Core of body
,18C
Maximum head
,38C
Maximum trunk
,39C
Maximum extremities
,40C
Specific absorption
Whole body (average)
,4 W/kg
rate (SAR)
Head (average)
,3 W/kg
Head or torso per gram
,8 W/kg
Extremities per gram
,12 W/kg
Peak pressure
200 pascals
Average pressure
105 dBA
Acoustic noise levels
rise time of gradients.

In some clinical applications, higher field strengths (e.g., 4.0 T) are allowed.
can occur. Because of risks to personnel, equipment, and physical facilities, manual
quenches should only be initiated after careful considerations and preparation.
Uncontrolled quenching is the result of a sudden loss of superconductivity in the
main magnet coils, which can result in the explosive conversion of liquid helium to
gas, and jeopardize the safety of those in the room and adjacent areas. In the event
of insufficient gas outflow, oxygen can be displaced and build up of pressure in the
room can prevent the entry door from being easily opened.
MRI is considered safe when used within the regulatory guidelines required
of the manufacturers by the Food and Drug Administration (Table 13-5). Serious
bioeffects are demonstrated with static and varying magnetic fields at strengths
significantly higher (10 to 20 times greater) than those used for typical diagnostic
imaging.
Static Magnetic Fields

The long-term biologic effects of high magnetic field strengths are not well known. At
lower magnetic field strengths, there have not been any reports of deleterious or nonreversable biologic effects, either acute or chronic. With very high field strength magnets
(e.g., 4 T or higher), there has been anecdotal mention of dizziness and disorientation
of personnel and patients as they move through the field. With systems in excess of 20
T, enzyme kinetic changes have been documented, increased membrane permeability
shown, and altered biopotentials have been measured. These effects have not been dem-
497
onstrated in magnetic fields below 10 T. Effects on ECG traces have shown an increased
amplitude of the T wave, presumably due to the magneto-hemodynamic effect, caused
by conductive fluid such as blood moving across a magnetic field. When this occurs,
sometimes the elevated T wave will be mistaken for the desired R wave, creating an
insufficient gating situation. For this reason, it is recommended that ECG leads are not
used for patient monitoring, but rather an alternative such as pulse oximetry be used.
Varying Magnetic Field Effects

The time-varying magnetic fields encountered in the MRI system are due to the gradient switching used for localization of the protons. Magnetic fields that vary their
strength with time are generally of greater concern than static fields because oscillating magnetic fields can induce electrical current flow in conductors. The maximum
allowed changing gradient fields depend on the rise times of the gradients, as listed
in Table 13-5. At extremely high levels of magnetic field variation, effects such as
visual phosphenes (the sensation of flashes of light being seen) can result because
of induced currents in the nerves or tissues. Other consequences such as bone healing and cardiac fibrillation have been suggested in the literature. The most common
bioeffect of MR systems is tissue heating caused by RF energy deposition and/or by
rapid switching of high strength gradients. RF coils and antennas can present burn
hazards when electrical currents and conductive loops are present, and must have
proper insulationboth electrical and thermal.
RF Exposure, Acoustic Noise Limits, Specific Absorption Rate

RF exposure causes heating of tissues. There are obvious effects of overheating, and
therefore a power deposition limit is imposed by governmental regulations on the
manufacturers for various aspects of MRI and MRS operation. Table 13-5 lists some
of the categories and the maximum values permitted for clinical use. The rationale for
imposing limits on static and varying magnetic fields is based on the ability of the resting body to dissipate heat buildup caused by the deposition and absorption of thermal energy. Other indicators, such as acoustic noise levels and pressure amplitudes,
are determined from limits that are shown to have reversible (unaffected) outcomes
for clinically approved sequences. Hearing protection should always be available to
the patient. For research sequences and procedures that have not been approved by
the FDA, patients or volunteers are to have hearing protection in place.
Pregnancy-related Issues and Pediatric Patient Concerns

Pregnant healthcare staff and physicians are permitted to work in and around the MR
environment throughout all stages of pregnancy and assist as needed in setting up
the patient and the exam. However, they are not to remain in the magnet scanning
room (Zone 4). Regarding the scanning of pregnant patients, current data have not
yielded any deleterious effects on the developing fetus with common examinations,
and no special considerations are warranted, as long as the benefit-risk-ratio of doing
the study is that for any other typical patient. Another consideration is the administration of contrast, which should not be routinely provided. A well-documented
consideration to administer contrast based on the overwhelming potential benefit to
the patient or fetus relative to the risk of exposing them to gadolinium-based agents
and potential deleterious effects of free gadolinium ions, as there are studies that have
documented that the agents do pass the placental barrier and do enter the fetus.
498
For pediatric patients, the largest issues are sedation and monitoring. Special
attention to sedation protocols, adherence to standards of care regarding sedation
guidelines, and monitoring patients during the scan with MR-safe temperature monitoring and isolation transport units are crucial to maintaining patient safety.
MR Personnel and MR Safety Zones

The American College of Radiology has published a white paper describing the recommendations for MR safety in general, and MR safety zones and MR personnel
definitions in particular in their 2007 publication (ACR white paper reference). MR
safety policies, procedures, and safe practices are suggested in the guidance. In an
effort to maintain a buffer zone around the always on magnet, MR safety zones
are categorized from Zone 1 to Zone 4. Zone 1 is the area freely accessible to the
general public, in essence everywhere outside of the MR magnet area and building.
Zone 2 represents the interface between Zone 1 and Zone 3typically the reception
area, where patients are registered and MR screening questions take place. Zone 3
is a restricted area comprised of the MR control room and computer room that only
specific personnel can access, namely, those specifically trained as MR personnel
(described below) and appropriately screened patients and other individuals (nurses,
FIGURE 13-47 Zoning concept for describing areas in an MR system environment include Zone 1, unrestricted access; Zone 2, interface between unrestricted and restricted areas; Zone 3, restricted area only allowed
for MR personnel and screened individuals; Zone 4, the area of the scanner and high magnetic field strength.
(Adapted from Kanal E., et.al. Guidance Document for Safe MR Practices: 2007.)
499
support staff). Zone 4 represents the MR magnet room, and is always located within
the confines of Zone 3. Demarcation of these zones should be clearly marked and
identified. Figure 13-47 illustrates the zoning concept.
Furthermore, there are personnel definitions describing criteria that must be
achieved before access can be granted to Zone 3 and Zone 4 areas. Non-MR personnel
are patients, visitors, for staff who do not have the appropriate education or training
that meet the criteria for Level 1 or Level 2 MR personnel. Level 1 MR personnel have
passed minimal safety and education training on MR safety issues and have a basic
understanding of the effects of MR magnets, dangers of projectiles, effects of strong
magnetic fields, etc. These individuals can work in Zone 3 and Zone 4 areas without
supervision or oversight; examples are MR office staff, patient aides, and custodial
staff. Level 2 MR personnel are more extensively trained in the broader aspects of MR
safety issues, for example, understanding the potential for thermal loading, burns,
neuromuscular excitation, and induced currents from gradients. These individuals
are the gatekeepers of access into Zone 4, and take the action and the leadership role
in the event of a patient code, ensuring that only properly screened support staff are
allowed access. Those responding to a code must be made aware of and comply with
MR safety protocols. Examples of personnel designated as Level 2 include MR technologists, MR medical physicists, radiologists, and department nursing staff.
Designated personnel are required to continuously maintain their safety credentials, by acquiring continuous education credits throughout the year on MR safety
aspects, and taking an annual test and achieving a minimum passing score.
13.10 Summary
Meeting the needs of the MR exam and using the equipment safely and effectively
requires the understanding of the basic physics underpinnings described in Chapter
12, and the details of advanced acquisition methods, image characteristics, artifacts/
pitfalls, and MR safety/bioeffects covered in this chapter. The reader is encouraged to
keep up with the rapid developments of MRI and MRS by referring to recent literature and websites dedicated to MRI education and technological advances.
REFERENCES AND SUGGESTED READINGS

American College of Radiology MRI Accreditation Program, Phantom Test Guidance and other
documentation for the MRI accreditation program, available at http://www.acr.org/accreditation/mri/
mri_qc_forms.aspx, accessed May 30, 2011.
Hendrick RE. Breast MRI: fundamentals and technical aspects. New York, NY: Springer, 2007.
Kanal E, Barkovich AJ, Bell C, et.al. ACR guidance document for safe MR practices: 2007. AJR Am J
Roentgenol. 2007;188:127.
NessAiver M. All you really need to know about MRI physics. Baltimore, MD: Simply Physics, 1997.
Al-Okaili RN, Krejza J, Wang S, Woo JH, Melhem ER. Advanced MR Imaging Techniques in the Diagnosis of
Intraaxial Brain Tumors in Adults. Radiographics 2006; 26: S173-S189.
Westbrook C, Kaut-Roth C, Talbot J. MRI in practice. 3rd ed. Malden, MA: Blackwell Publishing, 2005.
Chapter
14
Ultrasound
Ultrasound is the term that describes sound waves of frequencies exceeding the
range of human hearing and their propagation in a medium. Medical diagnostic
ultrasound is a modality that uses ultrasound energy and the acoustic properties
of the body to produce an image from stationary and moving tissues. In ultrasound imaging, a short pulse of mechanical energy is delivered to the tissues.
The pulse travels at the speed of sound, and with changes in the tissue acoustic properties, a fraction of the pulse is reflected as an echo that returns to the
source. Collection of the echoes over time and recording of the echo amplitudes
provide information about the tissues along the path of travel. Repeating the
process hundreds of times with a small incremental change in the direction of the
pulse interrogates a volume, from which a gray-scale tomographic image can be
synthesized. Generation of the sound pulses and detection of the echoes are accomplished with a transducer, which also directs the ultrasound pulse along a linear
path through the patient. Along a given beam path, the depth of an echo-producing
structure is determined from the time between the pulse emission and the echo
return, and the amplitude of the echo is encoded as a gray-scale value (Fig.14-1).
In addition to two-dimensional (2D) tomographic imaging, ultrasound provides
anatomical distance and volume measurements, motion studies, blood velocity
measurements, and 3D imaging.
Historically, medical uses of ultrasound came about shortly after the close of
World War II, derived from underwater sonar research. Initial clinical applications
monitored changes in the propagation of pulses through the brain to detect intracerebral hematoma and brain tumors based upon the displacement of the midline. Ultrasound rapidly progressed through the 1960s from simple A-mode scans
to B-mode applications and compound B-scan images using analog electronics. Advances in equipment design, data acquisition techniques, and data processing capabilities have led to electronic transducer arrays, digital electronics, and
real-time image display. This progress is changing the scope of ultrasound and its
applications in diagnostic radiology and other areas of medicine. High-resolution,
real-time imaging, harmonic imaging, 3D data acquisition, and power Doppler are
a few of the innovations introduced into clinical practice. Contrast agents for better
delineation of the anatomy, measurement of tissue perfusion, precise drug delivery
mechanisms, and determination of elastic properties of the tissues are examples of
current research.
The characteristics, properties, and production of ultrasound are described, followed by modes of ultrasound interaction, instrumentation, and image acquisition
methods. Signal processing and image display methods show how the ultrasound
image is produced, and how evaluation of the blood velocity through Doppler techniques is achieved. Types of image artifacts that can occur, quality control (QC) tests,
and potential ultrasound bioeffects complete the overview.
500
Chapter 14 Ultrasound
501
Transducer
Array
Interrogate body with

acoustic pulses generated
by transducer array
Acquire and record

echoes arising from
tissue interfaces
Construct
acoustic image
of tissues
FIGURE 14-1 A major use of ultrasound is the acquisition and display of the acoustic properties of tissues.
A transducer array (transmitter and receiver of ultrasound pulses) directs sound waves into the patient, receives
the returning echoes, and converts the echo amplitudes into a 2D tomographic image using the ultrasound
acquisition system. Exams requiring a high level of safety such as obstetrics are increasing the use of ultrasound
in diagnostic radiology. Photo credit: Emi Manning UC Davis Health System.
14.1 Characteristics of Sound
Propagation of Sound
Sound is mechanical energy that propagates through a continuous, elastic medium by
the compression and rarefaction of particles that comprise it. A simple model of an
elastic medium is that of a spring, as shown in Figure 14-2. Compression is caused
by a mechanical deformation induced by an external force (such as a plane piston),
with a resultant increase in the pressure of the medium. Rarefaction occurs following the compression eventas the backward motion of the piston reverses the force,
the compressed particles transfer their energy to adjacent particles with a subsequent
reduction in the local pressure amplitude. The section of the spring in contact with the
piston then stretches and becomes decompressed (a rarefaction), while the compression
continues to travel forward. With continued back-and-forth motion of the piston, a
Plane-piston
mechanical displacement
Elastic particles in the medium
t=1
t=2
t=3
t=4
Particle displacement
Compression
t=5
Rarefaction
+
Pressure
amplitude
variation
-
Time
FIGURE 14-2 Ultrasound energy

is generated by a mechanical displacement in compressible medium,
which is modeled as an elastic spring.
Energy propagation is shown as a
function of time, resulting in areas
of compression and rarefaction with
corresponding variations in positive
and negative pressure amplitude,
P (lower diagram). Particles in the
medium have only minor displacement as the wave front passes.
502
series of compressions and rarefactions propagate through the continuous medium (the
spring). Any change with respect to position in the elasticity of the spring disturbs the
wave train and causes some or all of the energy to be reflected. While the medium itself
is necessary for mechanical energy transfer (i.e., sound propagation), the constituent
particles of the medium act only to transfer mechanical energy; these particles experience only very small back-and-forth displacements. Energy propagation occurs as a
wave front in the direction of energy travel, known as a longitudinal wave.
Wavelength, Frequency, and Speed

The wavelength (l) of the ultrasound energy is the distance (usually expressed in
units of mm or m) between compressions or rarefactions, or between any two points
that repeat on the sinusoidal wave of pressure amplitude (see Fig. 14-2). The frequency (f) is the number of times the wave oscillates through one cycle each second
(s). Sound waves with frequencies less than 15 cycles per second (Hz) are called
infrasound, and the range between 15 Hz and 20 kHz comprises the audible acoustic
spectrum. Ultrasound represents the frequency range above 20 kHz. Medical ultrasound uses frequencies in the range of 2 to 10 MHz, with specialized ultrasound
applications up to 50 MHz. The period is the time duration of one wave cycle and is
equal to 1/f, where f is expressed in cycles/s. The speed of sound is the distance traveled by the wave per unit time and is equal to the wavelength divided by the period.
Since period and frequency are inversely related, the relationship between speed,
wavelength, and frequency for sound waves is
c lf
where c (m/s) is the speed of sound in the medium, l (m) is the wavelength, and
f (cycles/s) is the frequency.
The speed of sound is dependent on the propagation medium and varies widely
in different materials. The wave speed is determined by the ratio of the bulk modulus
(B) (a measure of the stiffness of a medium and its resistance to being compressed)
and the density (r) of the medium:
c=
B
r
SI units are kg/(m-s2), kg/m3, and m/s for B, r and c, respectively. A highly compressible medium, such as air, has a low speed of sound, while a less compressible medium, such as bone, has a higher speed of sound. A less dense medium has
a higher speed of sound than a denser medium (e.g., dry air versus humid air).
The speeds of sound in materials encountered in medical ultrasound are listed in
Table14-1. Of major importance is the average speed for soft tissue (1,540 m/s),
fatty tissue (1,450 m/s), and air (330 m/s). The difference in the speed of sound at
tissue boundaries is a fundamental property that generates echoes (and contrast)
in an ultrasound image. Medical ultrasound machines use a speed of sound of
1,540 m/s when determining localization of reflectors and creating the acoustic
image. The speed of sound in soft tissue can be expressed in different units; besides
1,540 m/s, most common are 154,000 cm/s and 1.54 mm/ms, which are often helpful
in simplifying conversion factors or estimating values.
The ultrasound frequency is unaffected by changes in sound speed as the acoustic beam propagates through different media. Thus, the ultrasound wavelength is
dependent on the medium. Examples of the wavelength change as a function of
frequency and propagation medium are given below.
503
TABLE 14-1DENSITY AND SPEED OF SOUND IN
TISSUES AND MATERIALS FOR MEDICAL

ULTRASOUND
MATERIAL
DENSITY (kg/m3)
c (m/s)
c (mm/ms)
Air
1.2
330
0.33
Lung
300
600
0.60
Fat
924
1,450
1.45
Water
1,000
1,480
1.48
Soft Tissue
1,050
1,540
1.54
Kidney
1,041
1,565
1.57
Blood
1,058
1,560
1.56
Liver
1,061
1,555
1.55
Muscle
1,068
1,600
1.60
Skull bone
1,912
4,080
4.08
PZT
7,500
4,000
4.00
Example: A 2-MHz beam has a wavelength in soft tissue of

l =
c
1540 m / s
mm
=
= 770 10-6 m = 7.7 10-4 m 1000
= 0.77 mm
f
2 106 / s
m
A 10-MHz ultrasound beam has a corresponding wavelength in soft tissue of

=
1540 m/ s
mm
= 154 10-6 m = 1.54 10-4 m 1000
0.15 mm
10 106 / s
m
Higher frequency sound has shorter wavelength, as shown in Figure 14-3.
Example: A 5-MHz beam travels from soft tissue into fat. Calculate the wavelength in each
medium, and determine the percent wavelength change.
In soft tissue,
l=
c 1540 m / s
=
= 3.08 10-6 m 0.31 mm
f
5 106 / s
In fat,
=
1450 m / s
= 2.9 10-6 m 0.29 mm
5 106 / s
A decrease in wavelength of 5.8% occurs in going from soft tissue into fat, due to the differences in the speed of sound. This is depicted in Figure 14-3 for a 5-MHz ultrasound beam.
The wavelength in mm in soft tissue can be calculated directly by dividing the

speed of sound in mm/s (c = 1,540 m/s = 1.54 mm/s) by the frequency in MHz,
as
l (mm, soft tissue) =
c 1.54 mm / ms 1.54 mm /106 s 1.54 mm

=
=
=
f
f ( MHz )
f (MHz)
f (106 / s)
504
FIGURE 14-3 Ultrasound wavelength is determined

by the frequency and the speed of sound in the propagation medium. Wavelengths in soft tissue are calculated for 2-, 5-, and 10-MHz ultrasound sources for
soft tissue (blue). A comparison of wavelength in fat
(purple) to soft tissue at 5 MHz is also shown.
l =0.77 mm
2 MHz
0.31 mm Soft Tissue
5 MHz
0.29 mm Fat
0.15 mm
10 MHz
0.2
0.4
0.6
0.8
1.0
Distance, mm
Thus, it can quickly be determined that the wavelength of a 10-MHz beam is

0.154 mm, and that of a 1-MHz beam is 1.54 mm by using this simple change in units.
The spatial resolution of the ultrasound image and the attenuation of the ultrasound beam energy depend on the wavelength and frequency, respectively. Ultrasound
wavelength affects the spatial resolution achievable along the direction of the beam. A
high-frequency ultrasound beam (small wavelength) provides better resolution and
image detail than a low-frequency beam; however, the depth of beam penetration is
significantly reduced at higher frequency. Ultrasound frequencies used for imaging
are therefore selected according to the imaging application. For body parts requiring greater travel distance of the sound waves (e.g., abdominal imaging), lower frequency ultrasound is used (3.5 to 5 MHz) to image structures at significant depths.
For small body parts or organs close to the skin surface (e.g., thyroid, breast), higher
frequency ultrasound is selected (7.5 to 10 MHz). Most medical imaging applications
use ultrasound frequencies in the range of 2 to 10 MHz.
Modern ultrasound equipment consists of multiple sound transmitters aligned in
an array that create sound beams independently. Interaction of two or more separate
ultrasound beams in a medium can result in constructive and/or destructive wave interference, as shown in Figure 14-4. The amount of constructive or destructive interference depends on several factors, but the most important are the phase (position of
the periodic wave with respect to a reference point) and amplitude of the interacting
beams. With phase differences occurring because of the distances between individual
transmitters and time of transmitter excitation, all of the individual beams interactions can generate complex interference patterns, which are very important in shaping and steering the composite ultrasound beam.
Pressure, Intensity, and the dB Scale

Sound energy causes particle displacements and variations in local pressure amplitude, P, in the propagation medium. Pressure amplitude is defined as the peak maximum or peak minimum value from the average pressure on the medium in the
absence of a sound wave. In the case of a symmetrical waveform, the positive and
505
Wave interference patterns

Input
A. Constructive
Output
B. Destructive
C. Complex
FIGURE 14-4 Interaction of waves in a propagation medium, plotted as amplitude versus time. A. Constructive interference occurs with two ultrasound waves of the same frequency and phase, resulting in a higher
amplitude output wave. B. Destructive interference occurs with the waves 180 degrees out-of-phase, resulting
in a lower amplitude output wave. C. Complex interference occurs when waves of slightly different frequency
interact, resulting in an output waveform of higher and lower amplitude.
negative pressure amplitudes are equal (lower portion of Fig. 14-2); however, in
most diagnostic ultrasound applications, the compressional amplitude significantly
exceeds the rarefactional amplitude. The SI unit of pressure is the Pascal (Pa), defined
as one Newton per square meter (N/m2). The average atmospheric pressure on earth
at sea level is approximately equal to 100,000 Pa. Diagnostic ultrasound beams typically deliver peak pressure levels that exceed ten times the earths atmospheric pressure, or about one MPa (mega Pascal).
Intensity, I, is the amount of power (energy per unit time) per unit area and is
proportional to the square of the pressure amplitude, I P2 . Medical diagnostic ultrasound intensity levels are described in units of milliwatts/cm2the amount
of energy per unit time per unit area. The absolute intensity level depends upon
the method of ultrasound production (e.g., pulsed or continuousa discussion of
these parameters can be found in Section 14.11 of this chapter). Relative intensity
and pressure levels are described as a logarithmic ratio, the decibel (dB). The relative
intensity in dB for intensity ratios is calculated as
I
relative intensity (dB) = 10 log 2
I1
where I1 and I2 are intensity values that are compared as a relative measure, and log
denotes the base 10 logarithm. In diagnostic ultrasound, the ratio of the intensity of the
incident pulse to that of the returning echo can span a range of one million times or
more. The logarithm function compresses the large and expands the small ratios into
a more manageable number range. An intensity ratio of 106 (e.g., an incident intensity
506
TABLE 14-2INTENSITY RATIO AND CORRESPONDING
DECIBEL VALUES
INTENSITY RATIO
DECIBELS (dB)
I2/I1
Log (I2/I1)
0.3
10
10
100
20
10,000
40
1,000,000
60
0.5
0.3
0.01
3
20
0.0001
40
0.000001
60
one million times greater than the returning echo intensity) is equal to 60 dB, whereas
an intensity ratio of 102 is equal to 20 dB. A change of 10 in the dB scale corresponds
to an order of magnitude (10 times) change in intensity; a change of 20 corresponds to
two orders of magnitude (100 times) change, and so forth. When the intensity ratio is
greater than one (e.g., the incident ultrasound intensity to the detected echo intensity),
the dB values are positive; when less than one, the dB values are negative. A loss of 3 dB
(3 dB) represents a 50% loss of signal intensity. The tissue thickness that reduces the
ultrasound intensity by 3 dB is considered the half-value thickness (HVT). In Table14-2,
intensity ratios, logarithms, and the corresponding intensity dB values are listed.
Example: Calculate the remaining intensity of a 100 mW ultrasound pulse that loses 30 dB
while traveling through tissue.
relative intensity (dB) = 10 log
I2
;
I1
I2
;
100 mW
I2
;
-3 = log
100 mW
I2
10-3 =
;
100 mW
-30 dB = 10 log
defining equation
divide each side of the equation by 10:
exponentiate using base 10
solve for I2 :
I2 = 0.001 100 mW = 0.1 mW
14.2 Interactions of Ultrasound with Matter

Ultrasound interactions are determined by the acoustic properties of matter. As ultrasound energy propagates through a medium, interactions include reflection, refraction, scattering, and absorption. Reflection occurs at tissue boundaries where there is
a difference in the acoustic impedance of adjacent materials. When the incident beam
is perpendicular to the boundary, a fraction of the beam (an echo) returns directly
507
back to the source; the transmitted fraction of the beam continues in the initial direction. For non-normal incidence, the incident angle is that made relative to normal
incidence; the reflected angle is equal to the incident angle. Refraction describes the
change in direction of the transmitted ultrasound energy with nonperpendicular
incidence. Scattering occurs by reflection or refraction, usually by small particles
within the tissue medium, causes the beam to diffuse in many directions, and gives
rise to the characteristic texture and gray scale in the acoustic image. Attenuation
refers to the loss of intensity of the ultrasound beam from absorption and scattering
in the medium. Absorption is the process whereby acoustic energy is converted to
heat energy, whereby, sound energy is lost and cannot be recovered.
Acoustic Impedance
The acoustic impedance (Z) of a material is defined as
Z = rc
where r is the density in kg/m3 and c is the speed of sound in m/s. The SI unit for
acoustic impedance is kg/(m2s), with the special name the rayl, where 1 rayl is equal
to 1 kg/(m2s). Table 14-3 lists the acoustic impedances of materials and tissues commonly encountered in medical ultrasonography. In a simplistic way, the acoustic
impedance can be likened to the stiffness and flexibility of a compressible medium
such as a spring, as mentioned at the beginning of this chapter. When springs with
different compressibility are connected together, the energy transfer from one spring
to another depends mainly on stiffness. A large difference in the stiffness results in a
large reflection of energy, an extreme example of which is a spring attached to a wall.
Minor differences in stiffness or compressibility allow the continued propagation of
energy, with little reflection at the interface. Sound propagating through a patient
behaves similarly. Soft tissue adjacent to air-filled lungs represents a large difference
in acoustic impedance; thus, ultrasonic energy incident on the lungs from soft tissue
is almost entirely reflected. When adjacent tissues have similar acoustic impedances,
only minor reflections of the incident energy occur. Acoustic impedance gives rise to
differences in transmission and reflection of ultrasound energy, which is the means
for producing an image using pulse-echo techniques.
TABLE 14-3ACOUSTIC IMPEDANCE, Z = c, FOR AIR,
WATER, AND SELECTED TISSUES

TISSUE
Z (RAYLS)
Air
0.0004 106
Lung
0.18 106
Fat
1.34 106
Water
1.48 106
Kidney
1.63 106
Blood
1.65 106
Liver
1.65 106
Muscle
1.71 106
Skull bone
7.8 106
508
Reflection
The reflection of ultrasound energy at a boundary between two tissues occurs
because of the differences in the acoustic impedances of the two tissues. The reflection coefficient describes the fraction of sound intensity incident on an interface that
is reflected. For perpendicular incidence (Fig. 14-5A), the intensity reflection coefficient, RI, is expressed as
Ir
Z - Z1
= 2
Z 2 + Z1
Ii
RI =
The subscripts 1 and 2 represent tissues proximal and distal to the boundary. The
intensity transmission coefficient, TI, is defined as the fraction of the incident intensity that is transmitted across an interface. With conservation of energy, the intensity
transmission coefficient is TI = 1 RI For a fat-muscle interface, the intensity reflection and transmission coefficients are calculated as
2
R I, (FatMuscle)
I r 1.71 1.34
=
0.015; TI, (FatMuscle) = 1 R I, (FatMuscle) = 0.985
I i 1.71 + 1.34
The actual intensity reflected or transmitted at a boundary is the product of the incident intensity and the reflection coefficient. For example, an intensity of 40 mW/cm2
incident on a boundary with RI = 0.015 reflects 40 0.015 = 0.6 mW/cm2. The
intensity reflection coefficient at a tissue interface is readily calculated from the acoustic impedance of each tissue. Examples of tissue interfaces and respective reflection
coefficients are listed in Table 14-4. For a typical muscle-fat interface, approximately
1% of the ultrasound intensity is reflected, and thus almost 99% of the intensity is
transmitted to greater depths in the tissues. At a muscle-air interface, nearly 100%
of incident intensity is reflected, making anatomy unobservable beyond an air-filled
cavity. This is why acoustic coupling gel must be used between the face of the transducer and the skin to eliminate air pockets. A conduit of tissue that allows ultrasound
transmission through structures such as the lung is known as an acoustic window.
Incident
qi
Z 1= r1c1
Reflection
(echo)
Z 2= r2c2
Transmission
c2 < c1
qr
qi
qr
Refraction
qt
c2 < c1
qt
FIGURE 14-5 Reflection and refraction of ultrasound occur at tissue boundaries with differences in acoustic impedance, Z. A. With perpendicular incidence (90 degrees), a fraction of the beam is transmitted and a
fraction of the beam is reflected to the source at a tissue boundary. B. With nonperpendicular incidence, (90
degrees), the reflected fraction of the beam is directed away from the transducer at an angle r i and the
transmitted fraction of the beam is refracted in the transmission medium at a transmitted refraction angle
greater than the incident angle (t . i) when c2 . c1. C. Refraction angle of the transmitted beam is less than
that of the incident angle when c1 > c2.
509
TABLE 14-4PRESSURE AND REFLECTION
COEFFICIENTS FOR VARIOUS INTERFACES

TISSUE INTERFACE
PRESSURE REFLECTION
INTENSITY REFLECTION
Liver-Kidney
0.006
0.00003
Liver-Fat
0.10
0.011
0.12
0.015
Fat-Muscle
Muscle-Bone
0.64
0.41
Muscle-Lung
0.81
0.65
Muscle-Air
0.99
0.99
When the beam is perpendicular to the tissue boundary, the sound is returned back
to the transducer as an echo. As sound travels from a medium of lower acoustic impedance into a medium of higher acoustic impedance, the reflected wave experiences a
180-degree phase shift in pressure amplitude, as shown by the negative sign on the
pressure amplitude values in Table 14-4. (Note: the reflected pressure amplitude is calculated as the ratio of the difference divided by the sum of the acoustic impedances.)
The above discussion assumes a smooth interface between tissues, where the
wavelength of the ultrasound beam is much greater than the structural variations of
the boundary. With higher frequency ultrasound beams, the wavelength becomes
smaller, and the boundary no longer appears smooth. In this case, returning echoes
are diffusely scattered throughout the medium, and only a small fraction of the incident intensity returns to the ultrasound transducer.
For nonperpendicular incidence at an angle i (Fig. 14-5B,C), the ultrasound
energy is reflected at an angle r equal to the incident angle, i = r Echoes are directed
away from the source of ultrasound and are thus undetected.
Refraction
Refraction describes the change in direction of the transmitted ultrasound energy at
a tissue boundary when the beam is not perpendicular to the boundary. Ultrasound
frequency does not change when propagating through various media, but a change
in the speed of sound may occur. Angles of incidence, reflection, and transmission
are measured relative to the normal incidence on the boundary, as shown in Figure
14-5B,C. The angle of refraction (t) is determined by the speed of sound change that
occurs at the boundary and is related to the angle of incidence (i) by Snells law:
sinqt
c
= 2
sinq i
c1
where i and t are the incident and transmitted angles, c1 and c2 are the speeds of
sound in medium 1 and 2, and medium 2 carries the transmitted ultrasound energy.
For small angles of incidence and transmission, Snells law can be approximated as
qt
c
2
qi
c1
When c2 > c1, the angle of transmission is greater than the angle of incidence as shown
in Figure 14-5B, and the opposite when c2<c1, as shown in Figure 14-5C. No refraction
510
occurs when the speed of sound is the same in the two media, or with perpendicular
incidence. This straight-line propagation is assumed in ultrasound signal processing,
and when refraction does occur, misplacement of anatomy in the image can result.
A situation called total reflection occurs when c2 > c1 and the angle of incidence of
the sound beam with the boundary between two media exceeds an angle called the
critical angle. In this case, the sound beam does not penetrate the second medium at
all but travels along the boundary. The critical angle (c) is calculated by setting t =
90 degree in Snells law (equation above), producing the equation sinq c = c1 c2.
Scattering
Acoustic scattering arises from objects and interfaces within a tissue that are about the
size of the wavelength or smaller and represent a rough or nonspecular reflector surface,
as shown in Figure 14-6. Most organs have a characteristic structure that gives rise to
a defined scatter signature and provides much of the diagnostic information contained in the ultrasound image. A specular reflector is a smooth boundary between two
media, where the dimensions of the boundary are much larger than the wavelength
of the incident ultrasound energy. Because nonspecular reflectors reflect sound in all
directions, the amplitudes of the returning echoes are significantly weaker than echoes
from tissue boundaries. Fortunately, the dynamic range of the ultrasound receiver is
sufficient to detect echo information over a wide range of amplitudes. Intensities of
returning echoes from nonspecular reflectors in the tissue parenchyma are not greatly
affected by beam direction, unlike the strong directional dependence of specular
reflectors. Parenchyma-generated echoes typically have similar echo strengths and
gray-scale levels in the image. Differences in scatter amplitude that occur from one
region to another cause corresponding brightness changes on the ultrasound display.
In general, the echo signal amplitude from the insonated tissues depends on the
number of scatterers per unit volume, the acoustic impedance differences at the scatterer interfaces, the sizes of the scatterers, and the ultrasonic frequency. Hyperechoic
(higher scatter amplitude) and hypoechoic (lower scatter amplitude) are terms used
for describing the scatter characteristics relative to the average background signal. Hyperechoic areas usually have greater numbers of scatterers, larger acoustic
impedance differences, and larger scatterers. Acoustic scattering from nonspecular
reflectors increases with frequency, while specular reflection is relatively independent
Tissue interactions:
Acoustic scattering
Small object reflectors

with size l
Boundary interactions:
Specular (smooth)
reflection
Non-specular
(diffuse) reflection
FIGURE 14-6 Ultrasound scattering. Small particle reflectors within a tissue or organ cause a diffuse scattering pattern that is characteristic of the particle size, giving rise to specific tissue or organ signatures. Boundary interactions with ultrasound can also scatter the beam, particularly at higher frequencies. Specular and
nonspecular boundary characteristics are partially dependent on the wavelength of the incident ultrasound.
As the wavelength becomes smaller, the boundary becomes rough, resulting in scattering from the surface
because of nonperpendicular boundaries.
511
TABLE 14-5ATTENUATION COEFFICIENTS FOR SELECTED
TISSUES AT 1 MHz
TISSUE COMPOSITION
ATTENUATION COEFFICIENT (1-MHz BEAM, dB/cm)
Water
0.0002
Blood
0.18
Soft tissues
0.30.8
Brain
0.30.5
Liver
0.40.7
Fat
0.51.8
Smooth muscle
0.20.6
Tendon
0.91.1
Bone, cortical
1326
Lung
40
of frequency. It is often possible to enhance the scattered echo signals over the specular echo signals by using higher ultrasound frequencies.
Attenuation
Ultrasound attenuation, the loss of acoustic energy with distance traveled, is caused
chiefly by scattering and tissue absorption of the incident beam. Absorbed acoustic energy is converted to heat in the tissue. The attenuation coefficient, , expressed
in units of dB/cm, is the relative intensity loss per centimeter of travel for a given
medium. Tissues and fluids have widely varying attenuation coefficients, as listed in
Table 14-5 for a 1-MHz ultrasound beam. Ultrasound attenuation expressed in dB
is approximately proportional to frequency. An approximate rule of thumb for soft
tissue is 0.5 dB per cm per MHz or 0.5 (dB/cm)/MHz. The product of the ultrasound
frequency (in MHz) with 0.5 (dB/cm)/MHz gives the approximate attenuation coefficient in dB/cm. Thus, a 2-MHz ultrasound beam will have approximately twice the
attenuation of a 1-MHz beam; a 10-MHz beam will have ten times the attenuation
per unit distance. Since the dB scale progresses logarithmically, the beam intensity
is exponentially attenuated with distance (Fig. 14-7). The ultrasound HVT is the
thickness of tissue necessary to attenuate the incident intensity by 50%, which is
equal to a 3 dB reduction in intensity. As the frequency increases, the HVT decreases,
as demonstrated by the examples below.
Example 1: Calculate the approximate intensity HVT in soft tissue for ultrasound beams of
2 and 10 MHz. Determine the number of HVTs the incident beam and the echo travel at a 6-cm
depth.
Answer: Information needed is (1) the attenuation coefficient approximation 0.5 (dB/cm)/MHz)
and (2) the HVT intensity expressed as a 3 dB loss. Given this information, the HVT in soft tissue
for a f MHz beam is
HVTf (MHz) (cm) =
3 dB
3 dB
6
=
=
attenuationcoefficient (dB / cm) 0.5(dB / cm) f (MHz) f
MHz
512
HVT2 MHz (cm) =
3 dB
= 3 cm
0.5 (dB /cm)
2 MHz
MHz
HVT10 MHz (cm) =
6
10
= 0.6 cm
Number of HVTs:
A 6-cm depth requires a travel distance of 12 cm (round-trip).
For a 2-MHz beam, this is 12 cm/(3 cm /HVT2 MHz) = 4 HVT2 MHz.
For a 10-MHz beam, this is 12 cm/(0.6 cm/HVT10 MHz) = 20 HVT10 MHz.
Example 2: Calculate the approximate intensity loss of a 5-MHz ultrasound wave traveling
round-trip to a depth of 4 cm in the liver and reflected from an encapsulated air pocket
(100% reflection at the boundary).
Answer: Using 0.5 dB/(cm-MHz) for a 5-MHz transducer, the attenuation coefficient is
2.5 dB/cm. The total distance traveled by the ultrasound pulse is 8 cm (4 cm to the depth of interest and 4cm back to the transducer). Thus, the total attenuation is 2.5 dB/cm 8 cm = 20 dB The
incident intensity relative to the returning intensity (100% reflection at the boundary) is
20 dB = 10 log(
2
= log(
102 =
I Incident
)
I Echo
I Incident
)
I Echo
I Incident
Therefore, I Incident = 100 I Echo
I Echo
The echo intensity is one-hundredth of the incident intensity in this example, or

20 dB. If the boundary reflected 1% of the incident intensity (a typical value), the
returning echo intensity would be (100/0.01) or 10,000 times less than the incident
intensity, or 40 dB. Considering the depth and travel distance of the ultrasound
energy, the detector system must have a dynamic range of 120 to 140 dB to be sensitive to acoustic signals generated in the medium. When penetration to deeper structures is important, lower frequency ultrasound transducers must be used. Another
Ultrasound Attenuation
1.0
Relative Intensity
FIGURE 14-7 Ultrasound attenuation occurs exponentially with penetration depth and increases with
increased frequency. The plots are estimates of a single frequency ultrasound
wave with an attenuation coefficient
of (0.5 dB/cm)/MHz of ultrasound
intensity versus penetration depth.
Note that the total distance traveled
by the ultrasound pulse and echo is
twice the penetration depth.
m=0.5 dB/cm
I = I0 e2mx
0.5
2 MHz
10 MHz
4 MHz
X
0
0
0.5
1.0
1.5
2.0 2.5
3.0 3.5
4.0
4.5
Penetration depth, cm
(Distance traveled = 2 penetration depth)
513
consequence of frequency-dependent attenuation is the preferential removal of the

highest frequency components in a broadband ultrasound pulse (discussed below in
Section14.3) and a shift to lower frequencies.
14.3 Ultrasound Transducers

Ultrasound is produced and detected with a transducer, comprised of one or more
ceramic elements with electromechanical properties and peripheral components.
The ceramic element converts electrical energy into mechanical energy to produce
ultrasound and mechanical energy into electrical energy for ultrasound detection.
Over the past several decades, the transducer assembly has evolved considerably in
design, function, and capability, from a single-element resonance crystal to a broadband transducer array of hundreds of individual elements. A simple single-element,
plane-piston source transducer is illustrated in Figure 14-8. Major components
include the piezoelectric material, matching layer, backing block, acoustic absorber,
insulating cover, tuning coil, sensor electrodes, and transducer housing.
Piezoelectric Materials
A piezoelectric material (often a crystal or ceramic) is the functional component of
the transducer. It converts electrical energy into mechanical (sound) energy by physical deformation of the crystal structure. Conversely, mechanical pressure applied
to its surface creates electrical energy. Piezoelectric materials are characterized by a
well-defined molecular arrangement of electrical dipoles, as shown in Figure 14-9.
Electrical dipoles are molecular entities containing positive and negative electric
charges that have an overall neutral charge. When mechanically compressed by
an externally applied pressure, the alignment of the dipoles is disturbed from the
equilibrium position to cause an imbalance of the charge distribution. A potential
difference (voltage) is created across the element with one surface maintaining a net
positive charge and one surface a net negative charge. Surface electrodes measure
the magnitude of voltage, which is proportional to the incident mechanical pressure amplitude. Conversely, application of an external voltage through conductors
attached to the surface electrodes induces the mechanical expansion and contraction
of the transducer element.
Piezoelectric materials are available from natural and synthetic materials. An example of a natural piezoelectric material is quartz crystal, commonly used in watches and
other timepieces to provide a mechanical vibration source at 32.768 kHz for interval
timing. (This is one of several oscillation frequencies of quartz, determined by the
To transmitter / receiver
Coaxial cable
Plastic Case
Tuning coil
Metal shield
Acoustic absorber
Backing block
Piezoelectric element
Matching layer(s)
FIGURE 14-8 A single-element ultrasound

transducer assembly is comprised of the piezoelectric ceramic, the backing block, acoustic absorber,
tuning coil and metal shield, transducer housing,
coaxial cable and voltage source, and the ceramic
to tissue matching layer.
514
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Compression
+
+
+
+
+
+
+
+
+
+
+
+
+
Equilibrium
Expansion
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Compression
Equilibrium:
No applied voltage
Expansion
FIGURE 14-9 The piezoelectric element is comprised of aligned molecular dipoles. A. Under the influence
of mechanical pressure from an adjacent medium (e.g., an ultrasound echo), the element thickness contracts
(at the peak pressure amplitude), achieves equilibrium (with no pressure), or expands (at the peak rarefactional
pressure) causing realignment of the electrical dipoles to produce positive and negative surface charge. Surface electrodes (not shown) measure the charge variation (voltage) as a function of time B. An external voltage
source applied to the element surfaces causes compression or expansion from equilibrium by realignment of
the dipoles in response to the electrical attraction or repulsion force.
crystal cut and machining properties.) Ultrasound transducers for medical imaging
applications employ a synthetic piezoelectric ceramic, most often lead-zirconate-titanate (PZT)a compound with the structure of molecular dipoles. The piezoelectric
attributes are attained after a process of molecular synthesis, heating, orientation of
internal dipole structures with an applied external voltage, cooling to permanently
maintain the dipole orientation, and cutting into a specific shape. For PZT in its
natural state, no piezoelectric properties are exhibited; however, heating the material
past its Curie temperature (e.g., 328C to 365C) and applying an external voltage
causes the dipoles to align in the ceramic. The external voltage is maintained until
the material has cooled to below its Curie temperature. Once the material has cooled,
the dipoles retain their alignment. At equilibrium, there is no net charge on ceramic

fo
FIGURE 14-10 A short-duration

voltage spike causes the resonance
piezoelectric element to vibrate at
its natural frequency, f0, which is
determined by the thickness of the
transducer equal to . Low-frequency oscillation is produced with
a thicker piezoelectric element. The
spatial pulse length (SPL) is a function of the operating frequency and
the adjacent damping block.
is determined by the transducer thickness equal to l
thick transducer
thin transducer
Pulse
Low frequency
515
Pulse
High frequency
surfaces. When compressed, an imbalance of charge produces a voltage between the

surfaces. Similarly, when a voltage is applied between electrodes attached to both surfaces, mechanical deformation occurs, as illustrated in Figure 14-9A,B.
Resonance Transducers
Resonance transducers for pulse-echo ultrasound imaging operate in a resonance
mode, whereby a voltage (usually 150 V) of very short duration (a voltage spike of
~1 s) is applied, causing the piezoelectric material to initially contract and then
subsequently vibrate at a natural resonance frequency. This frequency is selected
by the thickness cut of the PZT, due to the preferential emission of -wavelength
ultrasound waves in the piezoelectric material as illustrated in Figure 14-10. The
operating frequency is determined from the speed of sound in, and the thickness of,
the piezoelectric material. For example, a 5-MHz transducer will have a wavelength
in PZT (speed of sound in PZT is ~4,000 m/s) of
l =
c
4,000 m/s
=
f
5 106/s
= 8 104 m = 0.8 mm
To achieve the 5-MHz resonance frequency, a transducer element thickness of

0.8 mm = 0.4 mm is required. Higher frequencies are achieved with thinner elements,
and lower frequencies with thicker elements. Resonance transducers transmit and
receive preferentially at a single center frequency.
Damping Block
The damping block, layered on the back of the piezoelectric element, absorbs
the backward directed ultrasound energy and attenuates stray ultrasound signals
from the housing. This component also dampens the transducer vibration to create
an ultrasound pulse with a short spatial pulse length (SPL), which is necessary to
preserve detail along the beam axis (axial resolution). Dampening of the vibration
516
(also known as ring-down) lessens the purity of the resonance frequency and introduces a broadband frequency spectrum. With ring-down, an increase in the bandwidth (range of frequencies) of the ultrasound pulse occurs by introducing higher
and lower frequencies above and below the center (resonance) frequency. The Q
factor describes the bandwidth of the sound emanating from a transducer as
Q =
f0
bandwidth
where f0 is the center frequency and the bandwidth is the width of the frequency
distribution.
A high Q transducer has a narrow bandwidth (i.e., very little damping) and a
corresponding long SPL. A low Q transducer has a wide bandwidth and short SPL.
Imaging applications require a broad bandwidth transducer in order to achieve high
spatial resolution along the direction of beam travel. Blood velocity measurements by
Doppler instrumentation (explained in Section 14.9) require a relatively narrow-band
transducer response in order to preserve velocity information encoded by changes in
the echo frequency relative to the incident frequency. Continuous-wave ultrasound
transducers have a very high Q characteristic. An example of a high Q and low
Q ultrasound pulse illustrates the relationship to SPL in Figure14-11. While the Q
factor is derived from the term quality factor, a transducer with a low Q does not
imply poor quality in the signal.
Matching Layer
The matching layer provides the interface between the raw transducer element and
the tissue and minimizes the acoustic impedance differences between the transducer
and the patient. It consists of layers of materials with acoustic impedances that are
intermediate to soft tissue and the transducer material. The thickness of each layer
is equal to wavelength, determined from the center operating frequency of the
Light damping, high Q
Heavy damping, low Q

SPL short
SPL long
Damping block
Damping block
Frequency Spectrum
Amplitude
narrow bandwidth
broad bandwidth
1.0
1.0
0.5
0.5
0.0
0.0
0.8
0.9
1.0
f/fo
1.1
1.2
0.8
0.9
1.0
1.1
1.2
f/fo
FIGURE 14-11 Effect of damping block on the frequency spectrum. The damping block is adjacent to the
backside of the transducer and limits the vibration of the element to a small number of cycles. Light damping
allows many cycles to occur (top left), which results in an extended SPL (number of cycles times the wavelength) and a narrow frequency bandwidth (range of frequencies contained in the pulse (bottom left). Heavy
damping reduces the SPL and broadens the frequency bandwidth (top and bottom right). The Q factor
describes the center frequency divided by the bandwidth.
517
transducer and speed characteristics of the matching layer. For example, the wavelength of sound in a matching layer with a speed of sound of 2,000 m/s for a 5-MHz
ultrasound beam is 0.4 mm. The optimal matching layer thickness is equal to
= 0.4 mm = 0.1 mm. In addition to the matching layer, acoustic coupling gel
(with acoustic impedance similar to soft tissue) is used between the transducer and
the skin of the patient to eliminate air pockets that could attenuate and reflect the
ultrasound beam.
Nonresonance (Broad bandwidth) Multifrequency

Transducers
Modern transducer design coupled with digital signal processing enables multifrequency or multihertz transducer operation, whereby the center frequency can be
adjusted in the transmit mode. Unlike the resonance transducer design, the piezoelectric element is intricately machined into a large number of small rods and then
filled with an epoxy resin to create a smooth surface (Fig. 14-12). The acoustic properties are closer to tissue than a pure PZT material and thus provide a greater transmission efficiency of the ultrasound beam without resorting to multiple matching
layers. Broadband multifrequency transducers have bandwidths that exceed 80% of
the center frequency (Fig. 14-12B).
Multi-Hertz (Broadband) Transducer design
Native PZT material
Amplitude
Machined PZT rods
Broadband
80% Bandwidth
-40%
Epoxy backfill
Multi-Hertz
+40%
3 MHz
5 MHz
7 MHz
Center Frequency
3 MHz
7 MHz
5 MHz
Center Frequency
FIGURE 14-12 A. Multi-Hertz broadband transducer elements are created from a native piezoelectric
material into multiple small rods with epoxy backfill. This creates a ceramic element with acoustic characteristics closer to soft tissue and produces frequencies with very broad bandwidth. B. Bandwidth is often
described as a percentage of the center frequency. The graph shows an 80% bandwidth (40%) for a 5-MHz
frequency transducer (3- to 7-MHz operational sensitivity). C. Multi-Hertz operation depends on the broad
bandwidth transducer element to be sensitive to a range of returning frequencies during the reception of
echoes with subsequent digital signal processing to select the bandwidth(s) of interest.
518
Excitation of the multifrequency transducer is accomplished with a short square

wave burst of approximately 150 V with one to three cycles, unlike the voltage spike
used for resonance transducers. This allows the center frequency to be selected within
the limits of the transducer bandwidth. Likewise, the broad bandwidth response permits the reception of echoes within a wide range of frequencies. For instance, ultrasound pulses can be produced at a low frequency, and the echoes received at higher
frequency. Harmonic imaging is a recently introduced technique that uses this abilitylower frequency ultrasound is transmitted into the patient, and the higher frequency harmonics (e.g., two times the transmitted center frequency), created from the
interaction with contrast agents and tissues, are received as echoes. Native tissue harmonic imaging has certain advantages including greater depth of penetration, noise
and clutter removal, and improved lateral spatial resolution. Operational characteristics of multihertz transducers and harmonic imaging are explained in Section 14.7.
Transducer Arrays
The majority of ultrasound systems employ transducers with many individual rectangular piezoelectric elements arranged in linear or curvilinear arrays. Typically, 128 to
512 individual rectangular elements comprise the transducer assembly. Each element
has a width typically less than wavelength and a length of several millimeters. Two
modes of activation are used to produce a beam. These are the linear (sequential)
and phased activation/receive modes as illustrated in Figure 14-13.
Linear Arrays
Linear array transducers typically contain 256 to 512 elements; physically these are
the largest transducer assemblies. In operation, the simultaneous firing of a small
group of approximately 20 adjacent elements produces the ultrasound beam. The
simultaneous activation produces a synthetic aperture (effective transducer width)
defined by the number of active elements. Echoes are detected in the receive mode
by acquiring signals from most of the transducer elements. Subsequent A-line
(see Section 14.5) acquisition occurs by firing another group of transducer elements
displaced by one or two elements. A rectangular field of view (FOV) is produced with
this transducer arrangement. For a curvilinear array, a trapezoidal FOV is produced.
A
Linear Array
Phased Array
C Transducer
element
Height
Width
Thickness
A subset of transducer elements activated
All transducer elements activated
FIGURE 14-13 Multielement transducer arrays. A. A linear (or curvilinear) array produces a beam by firing
a subset of the total number of transducer elements as a group. B. A phased array produces a beam from
all of the transducer elements fired with fractional time delays in order to steer and focus the beam. C. The
transducer element in an array has a thickness, width, and height; the width is typically on the order of
wavelength; and the height depends on the transducer design and slice-thickness requirements.
519
Phased Arrays
A phased-array transducer is usually comprised of 64 to 128 individual elements in a
smaller package than a linear array transducer. All transducer elements are activated
nearly simultaneously to produce a single ultrasound beam. By using time delays
in the electrical activation of the discrete elements across the face of the transducer,
the ultrasound beam can be steered and focused electronically without physically
moving the transducer on the patient. During ultrasound signal reception, all of the
transducer elements detect the returning echoes from the beam path, and sophisticated detection algorithms synthesize the data to form the image.
Capacitive Micromachined Ultrasonic Transducers

Another method of producing high-frequency ultrasound is with the use of capacitive
micromachined ultrasound transducers (CMUT), which bring the fabrication technology of integrated circuits into the field of medical ultrasound. These devices, first
investigated in the early 1990s, are silicon-based electrostatic transducers, recently
shown to be competitive with the lead-zirconate-titanate for producing and receiving
ultrasonic data for patient imaging. The basic element of a CMUT is a capacitor cell
with a fixed electrode (backplate) and a free electrode (membrane). The principle
of operation is electrostatic transduction, whereby an alternating voltage is applied
between the membrane and the backplate, and the modulation of the electrostatic
force results in membrane vibration with the generation of ultrasound. Conversely,
when the membrane is subject to an incident ultrasound wave, the capacitance change
can be detected as a current or a voltage signal. For signal detection, a direct current
bias voltage must be used in the reception for signal detection and for transmission for linear operation of the array. Fabrication of these transducers with precisely
controlled geometric and mechanical properties is possible with microfabrication
technologies in the megahertz range; lateral dimensions of the membranes are on
the order of 10 microns and a thickness of about 1 to 2 m. Because of precise micromachining, the electrode separation is made small, which enables high electric fields
inside the gap and an ability to have high transduction efficiency.
The main advantages of CMUT arrays compared to PZT are better acoustic
matching with the propagation medium, which allows wider bandwidth capabilities,
improved resolution, potentially lower costs with easier fabrication, and the ability to have integrated circuits on the same wafer. CMUT arrays feature improved
axial resolution; however, further improvements in both sensitivity and resolution
are needed to fully compete with piezoelectric arrays, especially in areas where high
depth of penetration is required. Future 2D arrays show much promise in leading ultrasound to improvements in efficiency, speed, multibandwidth operation, and
volumetric imaging. These transducers will certainly have an impact on the ultrasound imaging systems of the future.
14.4 Ultrasound Beam Properties

The ultrasound beam propagates as a longitudinal wave from the transducer surface
into the propagation medium, and exhibits two distinct beam patterns: a slightly
converging beam out to a distance determined by the geometry and frequency of
the transducer (the near field), and a diverging beam beyond that point (the far
field). For an unfocused, single-element transducer, the distance of the near field is
determined by the transducer diameter and the frequency of the transmitted sound
520

A
Transducer element
Diameter, d
Length of near field

r2 / l = d2 / 4 l
Near field
Beam divergence
sin q = 1.22 l / d
Far field
Transducer
element
B
Pressure
Amplitude
Variations
FIGURE 14-14 A. The ultrasound beam from a single, circular transducer element is characterized by the
near field as a converging beam and far field as a diverging beam. B. Pressure amplitude variations in the near
field are quite complex and rapidly changing, while in the far field are monotonically decreasing.
(Fig. 14-14). For multiple transducer element arrays, an effective transducer diameter is determined by the excitation of a group of transducer elements. Because of
the interactions of each of the individual beams and the ability to focus and steer
the overall beam, the formulas for a single-element, unfocused transducer are not
directly applicable.
The Near Field

The near field, also known as the Fresnel zone, is adjacent to the transducer face and
has a converging beam profile. Beam convergence in the near field occurs because of
multiple constructive and destructive interference patterns of the ultrasound waves from
the transducer surface. Huygens principle describes a large transducer surface as an
infinite number of point sources of sound energy where each point is characterized as a
radial emitter. As individual wave patterns interact, the peaks and troughs from adjacent
sources constructively and destructively interfere causing the beam profile to be tightly
collimated in the near field. The ultrasound beam path is thus largely confined to the
dimensions of the active portion of the transducer surface, with the beam diameter converging to approximately half the transducer diameter at the end of the near field. The
near field length is dependent on the transducer diameter and propagation wavelength:
Near Field Length =
d2 r 2
= ;
4l l
where d is the transducer diameter, r is the transducer radius, and is the wave1.54 mm
length of ultrasound in the propagation medium. In soft tissue, l =
, and
f (MHz)
the near field length can be expressed as a function of diameter and frequency:
Near Field Length (soft tissue) =
d 2 (mm 2 ) f (MHz)
4 1.54 mm
This equation indicates that the near field distance is increased as the physical diameter and the operation frequency of the transducer are increased. Lateral resolution
521
(the ability of the system to resolve objects in a direction perpendicular to the beam
direction) is dependent on the beam diameter and is best at the end of the near field
for a single-element transducer. Lateral resolution is poor in areas close to and far
from the transducer surface.
Pressure amplitude characteristics in the near field are very complex, caused by
the constructive and destructive interference wave patterns of the ultrasound beam.
Peak ultrasound pressure occurs at the end of the near field, corresponding to the
minimum beam diameter for a single-element transducer. Pressures vary rapidly
from peak compression to peak rarefaction several times during transit through the
near field. Only when the far field is reached do the ultrasound pressure variations
decrease continuously (Fig. 14-14B).
The Far Field

The far field is also known as the Fraunhofer zone and is where the beam diverges.
For a large-area single-element transducer, the angle of ultrasound beam divergence,
, for the far field is given by
sin q = 1.22
l
d
where d is the effective diameter of the transducer and is the wavelength. Less beam
divergence occurs with high-frequency, large-diameter transducers. Unlike the near field,
where beam intensity varies from maximum to minimum to maximum in a converging
beam, ultrasound intensity in the far field decreases monotonically with distance.
Transducer Array Beam Formation and Focusing

In a transducer array, the narrow piezoelectric element width (typically between onehalf to one wavelength) produces a diverging beam at a distance very close to the
transducer face. Formation and convergence of the ultrasound beam occur with the
operation of several or all of the transducer elements at the same time. Transducer
elements in a linear array that are fired simultaneously produce an effective transducer width equal to the sum of the widths of the individual elements. Individual
beams interact via constructive and destructive interference to produce a collimated beam that has properties similar to the properties of a single transducer of the
same size. With a phased-array transducer, the beam is formed by interaction of the
individual wave fronts from each transducer, each with a slight difference in excitation time. Minor phase differences of adjacent beams form constructive and destructive wave summations that steer or focus the beam profile.
Transmit Focus
For a single transducer or group of simultaneously fired transducers in a linear array,
the focal distance is a function of the transducer diameter, the center operating frequency, and the presence of any acoustic lenses attached to the element surface. This
focal depth is unchangeable. Phased array transducers and many linear array transducers allow a selectable focal distance by applying specific timing delays between
transducer elements that cause the beam to converge at a specified distance. A shallow
focal zone (close to the transducer surface) is produced by firing outer transducers in
the array before the inner transducers in a symmetrical pattern, as shown in Figure
14-15. Greater focal distances are achieved by reducing the delay time differences
amongst the transducer elements, resulting in more distal beam convergence. Multiple
522
FIGURE 14-15 A phased-array transducer assembly uses all elements to produce the ultrasound beam.
Focusing is achieved by implementing a programmable delay time (beam former electronics) for the excitation
of the individual transducer elements (focusing requires the outer elements in the array be energized first).
Phase differences of the individual ultrasound pulses result in a minimum beam diameter (the focal distance)
at a predictable depth in tissue.
transmit focal zones are created by repeatedly acquiring data over the same volume,
but with different phase timing of the transducer array elements.
Receive Focus
In a phased-array transducer, the echoes received by all of the individual transducer
elements are summed together to create the ultrasound signal from a given depth.
Echoes received at the edge of the element array travel a slightly longer distance than
those received at the center of the array, particularly at shallow depths. Signals from
individual transducer elements therefore must be rephased to avoid a loss of resolution when the individual signals are synthesized to make an image. Dynamic receive
focusing is a method to rephase the signals by dynamically introducing electronic
delays as function of depth (time). At shallow depths, rephasing delays between adjacent transducer elements are greatest. With greater depth, there is less difference, so
the phase delay circuitry for the receiver varies as a function of echo listening time,
as shown in Figure 14-16. In addition to phased-array transducers, many linear array
transducers permit dynamic receive focusing amongst the active element group.
FIGURE 14-16 Dynamic receive focusing. All transducer elements in the phased
array are active during the receive mode,
and to maintain focus, the receive focus
timing must be continuously adjusted in
order to compensate for differences in
arrival time across the array as a function
of time (depth of the echo). Depicted
are an early time (top illustration) of
proximal echo arrival and a later time
of distal echo arrival. To achieve phase
alignment of the echo responses by all
elements, variable timing is implemented
as a function of element position after
the transmit pulse in the beam former.
The output of all phase-aligned echoes
is summed.
Dynamic Receive Focusing

Summed
Signal
Phase alignment
Proximal echoes
Variable timing circuitry

(length of bar indicates delay)
Distal echoes
Transducer crystal
front view
side view
Radial variations
Thickness variations
523
FIGURE 14-17 A. A single transducer of a

multielement array produces the main ultrasound beam in thickness mode vibration;
however, radial expansion and contraction
also occur. B. Side lobe energy in a multielement transducer array is created from the
radial transducer vibration, which is directed
away from the main beam. Echoes received
from the side lobe energy are mapped into
the main beam, creating unwanted artifacts.
Ultrasound beam
Side lobes
Main beam
Side lobes
Dynamic Aperture
The lateral spatial resolution of the linear array beam varies with depth, dependent on the linear dimension of the transducer width (aperture). A process termed
dynamic aperture increases the number of active receiving elements in the array
with reflector depth, so that the lateral resolution does not degrade with depth of
propagation.
Side Lobes and Grating Lobes

Side lobes are unwanted emissions of ultrasound energy directed away from the
main pulse, caused by the radial expansion and contraction of the transducer
element during thickness contraction and expansion (Fig. 14-17). In the receive
mode of transducer operation, echoes generated from the side lobes are unavoidably remapped along the main beam, which can introduce artifacts in the image.
In continuous mode operation, the narrow frequency bandwidth of the transducer (high Q) causes the side lobe energy to be a significant fraction of the
total beam. In pulsed mode operation, the low Q, broadband ultrasound beam
produces a spectrum of acoustic wavelengths that reduce the emission of side
lobe energy.
For multielement arrays, side lobe emission occurs in a forward direction along
the main beam. By keeping the individual transducer element widths small (less than
wavelength), the side lobe emissions are reduced. Another method to minimize
side lobes with array transducers is to reduce the amplitude of the peripheral transducer element excitations relative to the central element excitations.
Grating lobes result when ultrasound energy is emitted far off-axis by multielement arrays and are a consequence of the noncontinuous transducer surface of
the discrete elements. The grating lobe effect is equivalent to placing a grating in
front of a continuous transducer element, producing coherent waves directed at
a large angle away from the main beam (Fig. 14-18). This misdirected energy of
relatively low amplitude can result in the appearance of highly reflective objects
in the main beam.
524
FIGURE 14-18 Side and grating lobes are off-axis energy

emissions produced by linear and phased-array transducers. Side
lobes are forward directed; grating lobes are emitted from the
array surface at very large angles.
Grating
Lobes
Main Beam
Side
Lobes
Spatial Resolution
In ultrasound, the major factor that limits the spatial resolution and visibility of
detail is the volume of the acoustic pulse. The axial, lateral, and elevational (slicethickness) dimensions determine the minimal volume element (Fig. 14-19). Each
dimension has an effect on the resolvability of objects in the image.
Axial Resolution
Axial resolution (also known as linear, range, longitudinal, or depth resolution)
refers to the ability to discern two closely spaced objects in the direction of the
beam. Achieving good axial resolution requires that the returning echoes be distinct
without overlap. The minimal required separation distance between two reflectors is
Acoustic
Lens
Lateral
Slice thickness
(elevational)
Axial
Resolution components in 3-D space.
Elevational Profile of
Ultrasound Beam with depth
FIGURE 14-19 A. The axial, lateral, and elevational (slice-thickness) contributions in three dimensions are
shown for a phased-array transducer ultrasound beam. Axial resolution, along the direction of the beam, is
independent of depth; lateral resolution and elevational resolution are strongly depth dependent. Lateral
resolution is determined by transmit and receive focus electronics; elevational resolution is determined by the
height of the transducer elements. At the focal distance, axial is better than lateral and lateral is better than
elevational resolution. B. Elevational resolution profile with an acoustic lens across the transducer array
produces a focal zone in the slice-thickness direction.
Boundary separation
Spatial Pulse Length (SPL)
resolved
1
> SPL
unresolved
525
FIGURE 14-20 Axial resolution is equal to SPL.

Tissue boundaries that are separated by a distance
greater than SPL produce echoes from the first
boundary that are completely distinct from echoes
reflected from the second boundary, whereas
boundaries with less than SPL result in overlap
of the returning echoes. Higher frequencies reduce
the SPL and thus improve the axial resolution, as
shown in the lower diagram.
< SPL
Higher frequency
resolved
> SPL
one-half of the SPL to avoid the overlap of returning echoes, as the distance traveled
between two reflectors is twice the separation distance. Objects spaced closer than
SPL will not be resolved (Fig. 14-20).
The SPL is the number of cycles emitted per pulse by the transducer multiplied by the wavelength. Shorter pulses, producing better axial resolution, can be
achieved with greater damping of the transducer element (to reduce the pulse duration and number of cycles) or with higher frequency (to reduce wavelength). For
imaging applications, the ultrasound pulse typically consists of three cycles. At 5
MHz (wavelength of 0.31 mm), the SPL is about 3 0.31 = 0.93 mm, which provides an axial resolution of (0.93 mm) = 0.47 mm. At a given frequency, shorter
pulse lengths require heavy damping and low Q, broad bandwidth operation. For a
constant damping factor, higher frequencies (shorter wavelengths) give better axial
resolution, but the imaging depth is reduced. Axial resolution remains constant
with depth.
Lateral Resolution
Lateral resolution, also known as azimuthal resolution, refers to the ability to discern
as separate two closely spaced objects perpendicular to the beam direction. For both
single-element transducers and multielement array transducers, the beam diameter
determines the lateral resolution (see Fig. 14-21). Since the beam diameter varies
with distance from the transducer in the near and far field, the lateral resolution is
depth dependent. The best lateral resolution occurs at the near fieldfar field interface. At this depth, the effective beam diameter is approximately equal to the
transducer diameter. In the far field, the beam diverges and substantially reduces the
lateral resolution.
The lateral resolution of linear and curvilinear array transducers can be varied.
The number of elements simultaneously activated in a group defines an effective
transducer width that has similar behavior to a single-transducer element of the same
width. Transmit and receive focusing along a single line is possible to produce focal
zones with depth along each line.
For the phased-array transducer, focusing to a specific depth is achieved by both
beam steering and transmit/receive focusing to reduce the effective beam width and
improve lateral resolution, especially in the near field. Multiple transmit/receive
focal zones can be implemented to maintain lateral resolution as a function of depth
(Fig.14-22). Each focal zone requires separate pulse-echo sequences to acquire data.
One way to accomplish this is to acquire data along one beam direction multiple
526
Phased array
transducer
Lateral resolution
varies with depth
Point
targets
Beam
diameter
Image of
point targets
FIGURE 14-21 Lateral resolution indicates the ability to discern objects perpendicular to the direction of
beam travel and is determined by the beam diameter. Point objects in the beam are averaged over the effective beam diameter in the ultrasound image as a function of depth. Best lateral resolution occurs at the focal
distance; good resolution occurs over the focal zone.
times (depending on the number of transmit focal zones) and to accept only the
echoes within each focal zone, building up a single line of in-focus zones by meshing the information together. Increasing the number of focal zones improves overall
in-focus lateral resolution with depth, but the amount of time required to produce
an image increases, with a consequent reduction in frame rate and/or number of scan
lines per image (see section on image quality).
FIGURE 14-22 Phased-array transducers have multiple user

selectable transmit and receive focal zones implemented by
the beam former electronics. Each focal zone requires the
excitation of the entire array for a given focal distance. Subsequent processing meshes the independently acquired data to
enhance the lateral focal zone over a greater distance. Good
lateral resolution over an extended depth is achieved, but the
image frame rate is reduced.
Transmit
and
Receive
focusing
delays
Enhanced
Lateral
Focal Zone
Multiple transmit
focal distances
1.5 D array
527
FIGURE 14-23 Elevational resolution with

multiple transmit focusing zones is achieved
with 1.5D transducer arrays to reduce the
slice-thickness profile over an extended depth.
Five to seven discrete arrays replace the single
array. Phase delay timing provides focusing in
the elevational plane (similar to lateral transmit
and receive focusing).
Multiple transmit focal zones: elevational plane
Elevational resolution
Elevational Resolution
The elevational or slice-thickness dimension of the ultrasound beam is perpendicular to the image plane. Slice thickness plays a significant part in image resolution, particularly with respect to volume averaging of acoustic details in the
regions close to the transducer and in the far field beyond the focal zone. Elevational resolution is dependent on the transducer element height in much the
same way that the lateral resolution is dependent on the transducer element width
(Fig. 14-19B). Slice thickness is typically the weakest measure of resolution for
array transducers. Use of a fixed focal length lens across the entire surface of the
array provides improved elevational resolution at a fixed focal distance. Unfortunately, this compromises resolution due to partial volume averaging before and
after the elevational focal zone.
Multiple linear array transducers with five to seven rows, known as 1.5D transducer arrays, have the ability to steer and focus the beam in the elevational dimension. Elevational focusing is implemented with phased excitation of the outer to
inner arrays to minimize the slice-thickness dimension at a given depth (Fig. 14-23).
By using subsequent excitations with different focusing distances, multiple transmit
focusing can produce smaller slice thickness over a range of tissue depths. A disadvantage of elevational focusing is a frame rate reduction penalty required for multiple
excitations to build one image. The increased width of the transducer array also
limits positioning flexibility. Extension to future full 2D transducer arrays with
enhancements in computational power will allow 3D imaging with uniform resolution throughout the image volume.
14.5 Image Data Acquisition

Understanding ultrasonic image formation requires knowledge of ultrasound production, propagation, and interactions. Images are created using a pulse-echo mode
format of ultrasound production and detection. Each pulse is directionally transmitted into the patient and experiences partial reflections from tissue interfaces that
create echoes, which return to the transducer. Image formation using the pulse-echo
approach requires a number of hardware components: the beam former, pulser,
receiver, amplifier, scan converter/image memory, and display system (Fig. 14-24).
The detection and processing of the echo signals is the subject of this section.
528
Pulsers
Transmit/Receive
switches
Pre-amps
Beamformer:
Digital steering / focusing
Beam summation
A/Ds
Swept Gain
Patient
Image
Control
Panel
Receiver:
Time Gain
Log compression
Rectification
Rejection
Scan converter:
Memory
Digital interface
Post Processing
Storage
Transducers
Transducers
Tr
Transmit
mit gain
Keyboard
TGC
Acquisition and
display controls
B
FIGURE 14-24 A. Components of the ultrasound imager. This schematic depicts the design of a digital
acquisition/digital beam former system, where each of the transducer elements in the array has a pulser,
transmit-receive switch, preamplifier, and ADC (e.g., for a 128-element phased array, there are 128 components
as shaded boxes). Swept gain reduces the dynamic range of the signals prior to digitization. The beam former
provides focusing, steering, and summation of the beam; the receiver processes the data for optimal display,
and the scan converter produces the output image rendered on the monitor. Thick lines indicate the path of
ultrasound data through the system. B. A commercial ultrasound scanner system is comprised of a keyboard,
various acquisition and processing controls including transmit gain and TGC, several transducer selections, an
image display monitor, and other components/interfaces not shown.
529
Ultrasound equipment has digital electronics and processing, and current

state-of-the-art systems use various combinations of mostly digital and some analog electronics. The discussion below assumes a hybrid analog and digital processing capability for image data acquisition and refers to the modules illustrated in
Figure 14-24A.
Beam Former
The beam former is responsible for generating the electronic delays for individual
transducer elements in an array to achieve transmit and receive focusing and, in
phased arrays, beam steering. Most modern, high-end ultrasound equipment incorporates a digital beam former and digital electronics for both transmit and receive
functions. A digital beam former controls application-specific integrated circuits that
provide transmit/receive switches, digital-to-analog and analog-to-digital converters (ADCs), and preamplification and time gain compensation (TGC) circuitry for
each of the transducer elements in the array. Each of these components is explained
below. Major advantages of digital acquisition and processing include the flexibility
to introduce new ultrasound capabilities by programmable software algorithms and
enhanced control of the acoustic beam.
Pulser
The pulser (also known as the transmitter) provides the electrical voltage for exciting the piezoelectric transducer elements and controls the output transmit power by
adjustment of the applied voltage. In digital beam former systems, a digital-to-analog
converter (DAC) determines the amplitude of the voltage. An increase in transmit
amplitude creates higher intensity sound and improves echo detection from weaker
reflectors. A direct consequence is higher signal-to-noise ratio in the images but
also higher power deposition to the patient. User controls of the output power are
labeled output, power, dB, or transmit by the manufacturer. In some systems,
a low power setting for obstetric imaging is available to reduce power deposition
to the fetus. A method for indicating output power in terms of a thermal index
(TI) and mechanical index (MI) is provided by ultrasound equipment manufacturers
(see Section 14.11).
Transmit/Receive Switch
The transmit/receive switch, synchronized with the pulser, isolates the high voltage associated with pulsing (~150 V) from the sensitive amplification stages during
receive mode, with induced voltages ranging from approximately 1 V to 2 V from
the returning echoes. After the ring-down time, when vibration of the piezoelectric
material has stopped, the transducer electronics are switched to sensing surface
charge variations of mechanical deflection caused by the returning echoes, over a
period up to about 1,000 s (1 ms).
Pulse-Echo Operation
In the pulse-echo mode of transducer operation, the ultrasound beam is intermittently
transmitted, with a majority of the time occupied by listening for echoes. The ultrasound pulse is created with a short voltage waveform provided by the pulser of the
ultrasound system. This event is sometimes known as the main bang. The generated
530
pulse is typically two to three cycles long, dependent on the damping characteristics
of the transducer elements. With a speed of sound of 1,540 m/s (0.154 cm/s), the
time delay between the transmission pulse and the detection of the echo is directly
related to the depth of the interface as
Time ( ms) =
2D(cm )
2D(cm )
=
= 13ms / cm D(cm )
c (cm / ms) 0.154 cm / ms
Distance (cm ) =
c (cm /ms) Time (ms)

= 0.077 Time( ms)
2
where c, the speed of sound, is expressed in cm/s; distance from the transducer to
the reflector, D, is expressed in cm; the constant 2 represents the round-trip distance;
and time is expressed in s. One pulse-echo sequence produces one amplitudemodulated (A-line) of image data. The timing of the data excitation and echo acquisition relates to distance (Fig. 14-25). Many repetitions of the pulse-echo sequence
are necessary to construct an image from the individual A-lines.
The number of times the transducer is pulsed per second is known as the pulse
repetition frequency (PRF). For imaging, the PRF typically ranges from 2,000 to 4,000
pulses per second (2 to 4 kHz). The time between pulses is the pulse repetition period
(PRP), equal to the inverse of the PRF. An increase in PRF results in a decrease in echo
listening time. The maximum PRF is determined by the time required for echoes
from the most distant structures to reach the transducer. If a second pulse occurs
before the detection of the most distant echoes, these more distant echoes can be
confused with prompt echoes from the second pulse, and artifacts can occur. The
maximal range is determined from the product of the speed of sound and the PRP
divided by 2 (the factor of 2 accounts for round-trip distance):
Maximal range (cm ) = 154,000 cm / s PRP (s )
= 77,000 PRP (s ) = 77,000 / PRF( s 1 ).
A 500 s PRP corresponds to a PRF of 2 kHz and a maximal range of 38.5 cm. For
a PRP of 250 s (PRF of 4 kHz), the maximum depth is halved to 19.3 cm. Higher
ultrasound frequency operation has limited penetration depth, allowing high PRFs.
PRP
500 ms
Pulse
Duration
~1--2 msec
500 ms Pulse Repetition Period
pulse . .
pulse . . .
0 13
50
100
150
200
250
500
time, msec
0 1.0
3.9
7.7
11.6
15.4
19.3
38.5
range, cm
PRF =
1
1
1
2000
=
=
=
= 2 kHz
PRP 500 ms 500 10 -6 s
s
FIGURE 14-25 The pulse-echo timing of data acquisition depicts the initial pulse occurring in a very short
time span, the pulse duration = 1 to 2 s and the time between pulses, the PRP = 500 s in this example.
The number of pulses per second is 2,000/s, or 2 kHz. Range (one-half the round-trip distance) is calculated
assuming a speed of sound = 1,540 m/s.
531
onversely, lower frequency operation requires lower PRFs because echoes can return
C
from greater depths. Ultrasound transducer frequency should not be confused with
PRF, and the period of the sound wave (1/f) should not be confused with the PRP (1/
PRF). The ultrasound frequency is calibrated in MHz, whereas PRF is in kHz, and the
ultrasound period is measured in microseconds compared to milliseconds for the PRP.
Pulse duration is the ratio of the number of cycles in the pulse to the transducer
frequency and is equal to the instantaneous on time. A pulse consisting of two
cycles with a center frequency of 2 MHz has a duration of 1 s. Duty cycle, the
fraction of on time, is equal to the pulse duration divided by the PRP. For realtime imaging applications, the duty cycle is typically 0.2% to 0.4%, indicating that
greater than 99.5% of the scan time is spent listening to echoes as opposed to
producing acoustic energy. Intensity levels in medical ultrasonography are very low
when averaged over time, as is the intensity when averaged over space due to the
collimation of the beam.
For clinical data acquisition, a typical range of PRF, PRP, and duty cycle values
are listed in Table 14-6.
Preamplification and Analog-to-Digital Conversion

In multielement array transducers, all preprocessing steps are performed in parallel.
Each transducer element produces a small voltage proportional to the pressure
amplitude of the returning echoes. An initial preamplification increases the
detected voltages to useful signal levels. This is combined with a fixed swept gain
(Fig.14-26), to compensate for the exponential attenuation occurring with distance
(time) traveled. Large variations in echo amplitude (voltage produced in the piezoelectric element)with time are reduced from approximately 1,000,000:1 or 120 dB
to about 1,000:1 or 60 dB.
Early ultrasound units used analog electronic circuits for all functions, which
were susceptible to drift and instability. Even today, the initial stages of the receiver
often use analog electronic circuits. Digital electronics were first introduced in ultrasound for functions such as image formation and display. Since then, there has been
a tendency to implement more and more of the signal preprocessing functions in
digital circuitry, particularly in high-end ultrasound systems.
In state-of-the-art ultrasound units, each piezoelectric element has its own preamplifier and ADC. A typical sampling rate of 20 to 40 MHz with 8 to 12 bits of
precision is used (ADCs were discussed in Chapter 5). ADCs with larger bit depths
and sampling rates are necessary for systems that digitize the signals directly from the
preamplification stage. In systems where digitization of the signal occurs after analog
beam formation and summing, a single ADC with less demanding requirements is
typically employed.
TABLE 14-6TYPICAL PRF, PRP, AND DUTY CYCLE VALUES FOR
ULTRASOUND OPERATION MODES

OPERATION MODE
PRF (Hz)
PRP (s)
M-mode
500
2000
DUTY CYCLE (%)
0.05
Real-time
2000 4000
500 250
0.2 0.4
Pulsed Doppler
4000 12000
250 83
0.4 1.2
Adapted from Zagzebski, J. Essentials of ultrasound physics. St. Louis, MO: Mosby-Year Book, 1996.
532
FIGURE 14-26 A phased-array transducer produces a pulsed beam that is

focused at a programmable depth and
receives echoes during the PRP. This
figure shows a digital beam former system, with the analog to digital converter
(ADC) converting the signals before the
beam focus and steering manipulation.
Timed electronic delays phase align the
echoes, with the output summed to
form the ultrasound echo train along a
specific beam direction.
Returning
echoes
ADC
Digital dynamic
focusing
Summing
Circuit
Pre-amplification
and swept gain
Digital output
Time (Depth)
Beam Steering, Dynamic Focusing, and Signal Summation

Echo reception includes electronic delays to adjust for beam direction and dynamic
receive focusing to align the phase of detected echoes from the individual elements
in the array as a function of echo depth. In digital beam former systems, this is
accomplished with digital processing algorithms. Following phase alignment, the
preprocessed signals from all of the active transducer elements are summed as shown
in Figure 14-26. The output signal represents the acoustic information gathered during the PRP along a single beam direction. This information is sent to the receiver for
further processing before rendering into a 2D image.
Receiver
The receiver accepts data from the beam former during the PRP, which represents
echo information as a function of time (depth). Subsequent signal processing occurs
in the following sequence (Fig. 14-27):
1. Gain adjustments and dynamic frequency tuning.
TGC is a user-adjustable amplification of the returning echo signals as a function of time, to further compensate for beam attenuation. TGC (also known as
time varied gain, depth gain compensation, and swept gain) can be changed to
meet the needs of a specific imaging application. The ideal TGC curve makes all
equally reflective boundaries equal in signal amplitude, regardless of the depth
of the boundary (Fig. 14-28). Variations in the output signals are thus indicative
of the acoustic impedance differences between tissue boundaries. User adjustment is typically achieved by multiple slider potentiometers, where each slider
represents a given depth in the image, or by a 3-knob TGC control, which
controls the initial gain, slope, and far gain of the echo signals. For multielement transducers, TGC is applied simultaneously to the signal from each of the
individual elements. The TGC amplification effectively reduces the maximum
to minimum range of the echo voltages as a function of time to approximately
50 dB (300:1).
Dynamic frequency tuning is a feature of some broadband receivers that
changes the sensitivity of the tuner bandwidth with time, so that echoes from
shallow depths are tuned to a higher frequency range, while echoes from deeper
533
FIGURE 14-27 The receiver processes the data

streaming from the beam former. Steps include TGC,
dynamic range compression, rectification, demodulation, and noise rejection. The user has the ability to
adjust the TGC and the noise rejection level.
Detection
Pre-amplification
Receiver: (1) and (4) are user adjustable
(1) Time Gain Compensation
(2) Logarithmic Compression

(3) Demodulation and
Envelope detection
(4) Noise rejection level

(5) Processed signal
s tructures are tuned to lower frequencies. The purpose of this is to accommodate

for beam softening, where increased attenuation of higher frequencies in a broad
bandwidth pulse occurs as a function of depth. Dynamic frequency tuning allows
the receiver to make the most efficient use of the ultrasound frequencies incident
on the transducer.
Equally reflective acoustic impedance boundaries
After
TGC
Compression
Demodulation
Rejection
Amplitude
Gain
Amplitude
TGC
amplification
Exponential
Attenuation
Amplitude
Before
TGC
A-line
Time (depth)
FIGURE 14-28 TGC amplifies the acquired signals with respect to time after the initial pulse by operator
adjustments (usually a set of five to six slide potentiometers; see Fig. 14-24B). Processed A-line data with
appropriate TGC demonstrate that equally reflective interfaces have equal amplitude.
534
2. Dynamic range (logarithmic) compression. Dynamic range defines the effective

operational range of an electronic device from the threshold signal level to the
saturation level. Key components in the ultrasound detection and display that
are most affected by a wide dynamic range include the ADC and the display.
For receiver systems with an 8-bit ADC, the dynamic range is 20 log (256) =
48dB, certainly not enough to accurately digitize a 50-dB signal. Video monitors
and film have a dynamic range of about 100:1 to 150:1 and therefore require
a reduced range of input signals to accurately render the display output. Thus,
after TGC, the signals must be reduced to 20 to 30 dB, which is accomplished by
compression using logarithmic amplification to increase the smallest echo amplitudes and to decrease the largest amplitudes. Logarithmic amplification produces
an output signal proportional to the logarithm of the input signal. Logarithmic
amplification is performed by an analog signal processor in less costly ultrasound
systems and digitally in high-end digital systems having ADCs with a large bit
depth. In any case, an appropriate range of signals is achieved for display of the
amplitude variations as gray scale on the monitor or for printing on film.
3. Rectification, demodulation, and envelope detection. Rectification inverts
the negative amplitude signals of the echo to positive values. Demodulation and
envelope detection convert the rectified amplitudes of the echo into a smoothed,
single pulse.
4. Rejection level adjustment sets the threshold of signal amplitudes allowed to pass
to the digitization and display subsystems. This removes a significant amount of
undesirable low-level noise and clutter generated from scattered sound or by the
electronics.
5. Processed images are optimized for gray-scale range and viewing on the limited dynamic range monitors, so that subsequent adjustments to the images are
unnecessary.
In the above listed steps, the operator has the ability to control the TGC and noise/
clutter rejection. The amount of amplification (overall gain) necessary is dependent
on the initial power (transmit gain) settings of the ultrasound system. Higher intensities are achieved by exciting the transducer elements with larger voltages. This
increases the amplitude of the returning echoes and reduces the need for electronic
amplification gain but also deposits more energy into the patient, where heating or
mechanical interactions can be significant. Conversely, lower ultrasound power settings, such as those used in obstetrical ultrasound, require a greater overall electronic
gain to amplify weaker echo signals into the appropriate range for TGC. TGC allows
the operator to manipulate depth-dependent gain to improve image uniformity and
compensate for unusual imaging situations. Inappropriate adjustment of TGC can
lead to artifactual enhancement of tissue boundaries and tissue texture as well as
nonuniform response versus depth. The noise rejection level sets a threshold to clean
up low-level signals in the electronic signal. It is usually adjusted in conjunction with
the transmit power level setting of the ultrasound instrument.
Echo Display Modes

A-mode
A-mode (A for amplitude) is the display of the processed information from the receiver
versus time (after the receiver processing steps shown in Fig. 14-27). As echoes
return from tissue boundaries and scatterers (a function of the acoustic impedance
differences in the tissues), a digital signal proportional to echo amplitude is produced
535
as a function of time. One A-line of data per PRP is the result. As the speed of
sound equates to depth (round-trip time), the tissue interfaces along the path of the
ultrasound beam are localized by distance from the transducer. The earliest uses of
ultrasound in medicine used A-mode information to determine the midline position
of the brain for revealing possible mass effect of brain tumors. A-mode and A-line
information is currently used in ophthalmology applications for precise distance
measurements of the eye. Otherwise, A-mode display by itself is seldom used.
B-mode
B-mode (B for brightness) is the electronic conversion of the A-mode and A-line
information into brightness-modulated dots along the A-line trajectory. In general,
the brightness of the dot is proportional to the echo signal amplitude (depending
upon signal processing parameters). The B-mode display is used for M-mode and 2D
gray-scale imaging.
M-mode
M-mode (M for motion) is a technique that uses B-mode information to display the
echoes from a moving organ, such as the myocardium and valve leaflets, from a fixed
transducer position and beam direction on the patient (Fig. 14-29). The echo data
from a single ultrasound beam passing through moving anatomy are acquired and
displayed as a function of time, represented by reflector depth on the vertical axis
(beam path direction) and time on the horizontal axis. M-mode can provide excellent temporal resolution of motion patterns, allowing the evaluation of the function
of heart valves and other cardiac anatomy. Only one anatomical dimension is represented by the M-mode technique, and with advances in real-time 2D echocardiography, Doppler, and color flow imaging, this display mode is of much less importance
than in the past.
A-mode
M-mode
Incident
pulse
Returning
echoes
Valve Leaflets
B-mode
Time
Depth
Stationary
Transducer
Array
Resultant M-mode display
FIGURE 14-29 Ultrasonic data are acquired with a stationary transducer or transducer array. The A-mode
data (vertical blue trace) represent echo amplitudes during one pulse-echo period. The amplitudes are encoded
to brightness (B-mode) as a series of variable intensity dots. The A-mode line and corresponding B-mode dots
change position as the valve leaflets move within the stationary beam. By deflecting the dots horizontally in
time on a storage display, motion graphs (M-mode) are created, depicting the periodic (or lack of periodic)
motion.
536
Scan Converter
The function of the scan converter is to create 2D images from echo information from
distinct beam directions and to perform scan conversion to enable image data to be
viewed on video display monitors. Scan conversion is necessary because the image
acquisition and display occur in different formats. Early scan converters were of an
analog design, using storage cathode ray tubes to capture data. These devices drifted
easily and were unstable over time. Current scan converters use digital technology
for storage and manipulation of data. Digital scan converters are extremely stable and
allow subsequent image processing by the application of a variety of mathematical
functions.
Digital information streams to the scan converter memory, configured as a matrix
of small picture elements (pixels) that represent a rectangular coordinate display.
Most ultrasound instruments have an approximately 500 500 pixel matrix (variations between manufacturers exist). Each pixel has a memory address that uniquely
defines its location within the matrix. During image acquisition, the digital signals
are inserted into the matrix at memory addresses that correspond as close as possible to the relative reflector positions in the body. Transducer beam, orientation,
and echo delay times determine the correct pixel addresses (matrix coordinates) in
which to deposit the digital information. Misalignment between the digital image
matrix and the beam trajectory, particularly for sector-scan formats at larger depths,
requires data interpolation to fill in empty or partially filled pixels. The final image
is most often recorded with 512 512 8 bits per pixel, representing about
Mbytes of data. For color display, the bit depth is often as much as 24 bits (1byte
per primary color).
14.6 Two-Dimensional Image Display and Storage

A 2D ultrasound image is acquired by sweeping a pulsed ultrasound beam over
the volume of interest and displaying echo signals using B-mode conversion of
the A-mode signals. Echo position is based upon the delay time between the pulse
initiation and the reception of the echo, using the speed of sound in soft tissue
(1,540 m/s). The 2D image is progressively built up or continuously updated as
the beam is swept through the object.
Early B-mode ScannersManual Articulating Arm with

Static Display
Early B-mode scanners were made with a single-element transducer mounted on
an articulating arm with angular position encoders to determine the location of
the ultrasound beam path. This information was necessary to place echo signals at
proper positions in the image. The mechanical arm constrained the transducer to a
plane, so that the resultant image depicted a tomographic slice. An image was built
up on an analog scan converter and storage display by repeated pulsing and positional changes of the transducer (Fig. 14-30), requiring several seconds per image.
Linear, sector, or compound scans could be performed. Compound scans used a
combination of linear and sector transducer motions to improve the probability of
perpendicular pulse incidence to organ boundaries so that echoes would return to
the transducer. Image quality with these systems was highly dependent on the skill
of the sonographer.
537
Articulating Arm Single Transducer System

Position encoders
Scan conversion and

B-Mode image display:
compound scanning
3
3
2
2
1
1
Transducer
FIGURE 14-30 Articulating arm B-mode scanning produces an acoustic tomographic slice of the body.
Position encoders track the 3D position of the transducer to allow the scan converter to place the gray-scale
encoded A-line data in the 2D image plane. Shown is a compound scan (multiple compound directions of the
single-element transducer) along a plane of the body and the corresponding image lines. Compounding was
performed manually, building an image over several seconds, to ensure that the ultrasound beam was perpendicular to the boundary at some point during the scan.
Mechanical Scanning and Real-Time Display

The next step in the evolution of medical ultrasound imaging was dynamic scanning
with real-time display. This was achieved by implementing periodic mechanical
motion of the transducer. Early scanners used a single-element transducer to produce a real-time sector scan image by wobbling the transducer. Enhanced performance was obtained with rotation of a number of transducers on a wheel within the
transducer housing. The update of the screen display was determined by the PRF, the
oscillation or rotation frequency of the transducer, and the positional information
provided by an encoder attached to the rocking or rotating transducer. The number
of A-lines that comprised the image was determined by the penetration depth and
the image update rate.
Electronic Scanning and Real-Time Display

State-of-the-art ultrasound scanners employ array transducers with multiple piezoelectric elements to electronically sweep an ultrasound beam across the volume of
interest for dynamic ultrasound imaging. Array transducers are available as linear/
curvilinear arrays and phased arrays. They are distinguished by the way in which the
beam is produced and by the FOV coverage that is possible.
Linear and curvilinear array transducers produce rectangular and trapezoidal
images, respectively (Fig. 14-31). They are typically composed of 256 to 512 discrete
transducer elements of to 1 wavelength width each in an enclosure from about
6 to 8 cm wide. A small group of adjacent elements (~15 to 20) is simultaneously activated to create an active transducer area defined by the width (sum of the individual
element widths in the group) and the height of the elements. This beam propagates
perpendicular to the surface of the transducer, with a single line of echo information
acquired during the PRP. A shift of one or more transducer elements and repeating the simultaneous excitation of the group produce the next A-line of data. The
ultrasound beam sweeps across the volume of interest in a sequential fashion, with
the number of A-lines approximately equal to the number of transducer elements.
Advantages of the linear array are the wide FOV for regions close to the transducer
and uniform, rectangular sampling across the image. Electronic delays within the
538
Linear Array
Convex Curvilinear Array:
Rectangular Field of View
Trapezoidal Field of View
Ultrasound
beam
One A line of data
A lines
FIGURE 14-31 Linear and curvilinear array transducers produce an image by activating a subgroup of the
transducer elements that form one A-line of data in the scanned object and shifting the active elements by one
to acquire the next line of data. Linear arrays produce rectangular image formats; curvilinear arrays produce a
trapezoidal format with a wide FOV.
subgroup of transducer elements allow transmit and dynamic receive focusing for
improved lateral resolution with depth.
Phased-array transducers are typically comprised of a tightly grouped array of
64, 128, or 256 transducer elements in a 3- to 5-cm-wide enclosure. All transducer
elements are involved in producing the ultrasound beam and recording the returning echoes. The ultrasound beam is steered by adjusting the delays applied to the
individual transducer elements by the beam former. This time delay sequence is
varied from one transmit pulse to the next in order to change the sweep angle across
the FOV in a sector scan. (Fig. 14-32 shows three separate beams during the sweep.)
Beam steering
Time
Transmit / Receive Timing Delay
Composite beam area
FIGURE 14-32 The phased-array transducer electronically steers the ultrasound beam by introducing phase
delays during the transmit and receive timing of the beam former. Lateral focusing also occurs along the beam
direction. A sector format composite image is produced (right), with the number of A-lines dependent on
several imaging factors discussed in the text.
539
A similar time-delay strategy (dynamic receive focusing) is used to spatially

synchronize the returning ultrasound echoes as they strike each transducer element
(see Fig. 14-26). In addition to the beam steering capabilities, lateral focusing,
multiple transmit focal zones, and dynamic receive focusing are used in phased
arrays.
The small overall size of the phased-array transducer entrance window allows
flexibility in positioning, particularly through the intercostal spacea small ultrasound conduit in front of the lungsto allow cardiac imaging. In addition, it is
possible to use the phased-array transducer in conjunction with external triggers,
such as EKG, to produce M-mode scans at the same time as 2D images, and to allow
duplex Doppler/color flow imaging.
Spatial compounding is a method in which ultrasound information is obtained
from several different angles of insonation and combined to produce a single image.
In fact, the idea of spatial compound scans existed from the earliest implementations
of static ultrasound B-scan imaging, as depicted in Figure 14-30. In linear array
transducer systems, electronic beam steering allows the insonation of tissues from
multiple angles as shown in Figure 14-33, and by averaging the data, the resultant
compound image improves image quality in a variety of applications including breast
imaging, thyroid, atherosclerotic plaque, and musculoskeletal ultrasound imaging.
As each image is produced from multiple angles of insonation, the likelihood that
one of these angles will be perpendicular to a specular reflector is increased, and
in turn higher echo amplitudes are generated for better definition in the image. In
addition, curved surfaces appear more continuous. Speckle noise, a random source
of image variation, is reduced by the averaging process of forming the compound
image, with a corresponding increase in signal-to-noise ratio. One downside to spatial compounding is the persistence effect of frame averaging, the loss of temporal
resolution, and the increase in spatial blurring of moving objects, so it is not particularly useful in situations with voluntary and involuntary patient motion. Spatial
compounding is also implemented on phased-array transducers.
Normal scan
Spatial
Compounding
FIGURE 14-33 A vector phased array transducer (left) in the normal acquisition mode insonates tissues to
yield a sector image. Compound scanning (right) uses ultrasound beams produced at several angles achieved
by electronic steering (three are shown above) to acquire directional acoustic image data subsequently averaged to produce a single image. Note the oversampling patterns in this mode, and since the scan lines are
acquired sequentially, the frame rate is reduced by the number of insonation angles used to produce the
compound image.
540
Real-Time Ultrasound Imaging: Frame Rate, FOV, Depth,

and Spatial Sampling Trade-Offs
The 2D image (a single frame) is created from a number of A-lines, N (typically 100 or
more), acquired across the FOV. A larger number of lines will produce a higher quality image; however, the finite time for pulse-echo propagation places an upper limit
on N that also impacts the desired temporal resolution. The acquisition time for each
line, Tline = 13 ms/cm D (cm), is required for the echo data to be unambiguously
collected from a depth, D (see Fig. 14-25, which illustrates time and depth relationships). Thus, the time necessary per frame, Tframe, is given by N Tline = N 13 ms/cm
D (cm). The frame rate per second is the reciprocal of the time required per frame:
Frame rate =
1
Tframe
1
0.077/ms
77000 / s
=
=
N 13 ms D(cm) N D(cm)
N D(cm)
This equation describes the maximum frame rate possible in terms of N and D. If either
N or D increases without a corresponding decrease of the other variable, then the maximum frame rate will decrease. For a given procedure, the sonographer must consider
the compromises among frame rate, imaging depth, and number of lines/frame. A
secondary consideration is the line density (LD) (spacing between lines), determined
by N and the FOV. Higher frame rates can be achieved by reducing the imaging depth,
number of lines, or FOV as illustrated in Figure 14-34. The spatial sampling (LD) of
the ultrasound beam decreases with depth for sector and trapezoidal scan formats and
remains constant with depth for the rectangular format (linear array).
Low frame rate
High frame rate

= FOV
=N
= LD
Depth
Field of View:
Number of lines:
Line Density:
Penetration Depth:
FOV
N
LD
D
= FOV
N
LD
= Depth
FOV
FOV
N
= LD
= Depth
FIGURE 14-34 Ultrasound image quality depends on several factors. The number of lines per image, frame
rate, field of view (FOV), line density (LD), and penetration depth (D) are interrelated (see inset figure). For
a low frame rate image acquisition with a large penetration depth, a large number of A-lines and a high
LD, depicted on the right, are options that can be made with a high frame rate acquisition. Reducing the
penetration depth allows an equal LD, equal number of lines, and equal FOV (top right). Maintaining equal
penetration depth and equal FOV reduces the number of lines and the LD (middle right). Keeping the LD and
penetration depth equal reduces the FOV and number of lines (bottom right).
541
Another factor that affects frame rate is transmit focusing, whereby the ultrasound
beam (each A-line) is focused at multiple depths for improved lateral resolution (see
Fig. 14-22). The frame rate will be decreased by a factor approximately equal to the
number of transmit focal zones placed on the image, since the beam former electronics must transmit an independent set of pulses for each focal zone.
Scan LD is an important component of image quality. Insufficient LD can cause
the image to appear pixelated from the interpolation of several pixels to fill unscanned
volumes and can cause the loss of image resolution, particularly at a depth. This
might happen when one achieves high temporal resolution (high frame rates) at the
expense of LD.
Image Display
For digital flat-panel displays, digital information from the scan converter can be
directly converted into a viewable image. For analog monitor displays, the digital
scan converter memory requires a DAC and electronics to produce a compatible
video signal. The DAC converts the digital pixel data, in a raster pattern, into a corresponding analog video signal compatible with specific video monitors (refer to
Chapter 5 for detailed explanation). Window and level adjustments of the digital
data, to modify the brightness and contrast of the displayed image, are applied to
the output signals without changing the digital memory values (only the look-uptransformation table), before digital-to-analog conversion.
The displayed pixel density can limit the quality and resolution of the image. Employing a zoom feature on many ultrasound instruments can enhance the image information to improve and delineate details within the image that are otherwise blurred. Two
types of methods, read zoom and write zoom, are usually available. Read zoom
enlarges a user-defined region of the stored image and expands the information over
a larger number of pixels in the displayed image. Even though the displayed region
becomes larger, the resolution of the image itself does not change. Using write zoom
requires the operator to rescan the area of the patient that corresponds to the userselectable area. When enabled, the transducer scans the selected area and only the echo
data within the limited region are acquired. Figure 14-35 demonstrates the large FOV
image and a write zoom resampling of the image content. The latter allows a greater LD,
and higher sampling across the FOV provides improved resolution and image quality.
Besides the B-mode data used for the 2D image, other information from M-mode
and Doppler signal processing can also be displayed. During operation of the ultrasound scanner, information in the memory is continuously updated in real time.
When ultrasound scanning is stopped, the last image acquired is displayed on the
screen until ultrasound scanning resumes.
FIGURE 14-35 Large FOV sector scan image (left) and corresponding write zoom image (right).
542
Image Storage
Ultrasound images are typically comprised of 640 480 or 512 512 pixels. Each
pixel has a depth of 8 bits (1 byte) of digital data, providing up to 256 levels of gray
scale. Image storage (without compression) is approximately MB per image. For realtime imaging (10 to 30 frames per second), this can amount to hundreds of megabytes
of data for video clip acquisitions, depending on the complexity of the examination.
Color images used for Doppler studies (Section 14.7) increase the storage requirements
further because of larger numbers of bits needed for color resolution (full fidelity color
requires 24 bits/pixel, one byte each for the red, green, and blue primary colors).
14.7 Doppler Ultrasound

Doppler ultrasound is based on the shift of frequency in an ultrasound wave caused
by a moving reflector, such as blood cells in the vasculature (Fig. 14-36). This is
the same effect that causes a siren on a fire truck to sound high pitched as the truck
approaches the listener (the wavelength is compressed) and a shift to a lower pitch
sound as it passes by and continues on (the wavelength is expanded). The moving
reflectors in the body are the blood cells. By comparing the incident ultrasound frequency with the reflected ultrasound frequency from the blood cells, it is possible to
discern the velocity of the blood. Not only can blood velocity (and indirectly blood
flow) be measured, but the information provided by the Doppler techniques can also
be used to create color blood flow maps of the vasculature. The interpretation of
Doppler signals in clinical practice, however, requires the extraction of information
about the blood flow from the potential confounding aspects related to the technique
itself. Therefore, an understanding of the physical principles of Doppler ultrasound
is an absolute prerequisite for the interpretation of the acquired information.
Doppler Frequency Shift

The Doppler shift is the difference between the incident frequency and reflected frequency. When the reflector is moving directly away from or toward the source of
sound, the Doppler frequency shift (fd) is calculated as
fd = fr fi =
reflector speed
2 fi
reflector speed + speed of sound
FIGURE 14-36 Doppler ultrasound exploits changes in

frequency from interaction with moving objects. Sound
waves reflected from a moving object are compressed
(higher frequency) when moving toward the transducer and
expanded (lower frequency) when moving away from the
transducer compared to the incident sound wave frequency.
The difference between the incident and returning frequencies is called the Doppler shift frequency.
Blood moving towards transducer

produces higher frequency echoes.
Blood moving away from transducer

produces lower frequency echoes.
Measured Doppler shift

for velocity at angle q
Doppler angle,
543
FIGURE 14-37 Geometry of Doppler ultrasound data acquisition. The Doppler shift varies
as a function of the angle () of the incident ultrasound pulse and the axis of the blood vessel for
a fixed blood velocity, v. The maximum Doppler
shift occurs at an angle 0 For a larger Doppler
angle (), the measured Doppler shift is less by a
factor of cos(), and velocity estimates are compensated by 1/cos ().
blood cells:
Velocity, v m/s
Doppler shift for velocity
at q = 0
where fI is the frequency of the sound incident on the reflector and fr is the frequency
of the reflected sound. Thus, the Doppler shift is proportional to the velocity of the
blood cells.
When the sound waves and blood cells are not moving in parallel directions, the
equation must be modified to account for less Doppler shift. The angle between the
direction of blood flow and the direction of the sound is called the Doppler angle
(Fig. 14-37). The component of the velocity vector directed toward the transducer
is less than the velocity vector along the vessel axis by the cosine of the angle, cos
(). Without correction for this discrepancy, the Doppler shift will be less and an
underestimate of the actual blood velocity will occur. The Doppler angle joins the
adjacent side and hypotenuse of a right triangle; therefore, the component of the
blood velocity in the direction of the sound (adjacent side) is equal to the actual
blood velocity (hypotenuse) multiplied by the cosine of the angle (). (Note: In
physics, the term velocity refers to a vector quantity, describing both the distance
traveled per unit time (speed) and the direction of movement such as blood flow.) As
the velocity of blood cells (peak of ~200 cm/s) is significantly less than the speed of
sound (154,000 cm/s), the denominator can be simplified with an extremely small
error by neglecting the velocity of the blood. This results in the generalized Doppler
shift equation:
fd =
2 fi v cos(q )
c
where v is the velocity of blood, c is the speed of sound in soft tissue, and is the
Doppler angle.
Calculation of the blood velocity is straightforward by rearranging the Doppler
equation:
v=
fd c
2 fi cos(q )
Thus, the measured Doppler shift at a Doppler angle is adjusted by 1/cos() in

order to achieve accurate velocity estimates. Selected cosine values are cos 0 = 1, cos
30 = 0.87, cos 45 = 0.707, cos 60 = 0.5, and cos 90 = 0. At a 60-degree Doppler
angle, the measured Doppler frequency is the actual Doppler frequency, and at
90 degrees, the measured frequency is 0. The preferred Doppler angle ranges from
30 to 60 degrees. At too large an angle (greater than 60 degrees), the apparent
544
TABLE 14-7DOPPLER ANGLE AND ERROR ESTIMATES OF BLOOD
VELOCITY FOR A A +3-DEGREE ANGLE ACCURACY ERROR

ANGLE
(DEGREES)
SET ANGLE
(DEGREES)
ACTUAL VELOCITY
(cm/s)
ESTIMATED
VELOCITY (cm/s)
PERCENT
ERROR (%)
100
100.1
0.14
25
28
100
102.6
2.65
45
48
100
105.7
5.68
60
63
100
110.1
10.1
80
83
100
142.5
42.5
oppler shift is small, and minor errors in angle accuracy can result in large errors in
D
velocity (Table 14-7). At too small an angle (e.g., less than 20 degrees), refraction and
critical angle interactions can cause problems, as can aliasing of the signal in pulsed
Doppler studies.
The Doppler frequency shifts for moving blood occur in the audible range. It is
both customary and convenient to convert these frequency shifts into an audible signal through a loudspeaker that can be heard by the sonographer to aid in positioning
and to assist in diagnosis.
Example Given: f i = 5 MHz, v = 35 cm/s, and = 45 degrees, calculate the Doppler shift
frequency.
fd =
2 5 106 / s 35 cm / s 0.707
= 1.6 103 / s = 1.6 kHz
154,000 cm / s
The frequency shift of 1.6 kHz is in the audible range (15 Hz to 20 kHz). With an increased Doppler angle, the measured Doppler shift is decreased according to cos () since the projection of
the velocity vector toward the transducer decreases (and thus the audible frequencies decrease to
lower pitch).
Continuous Doppler Operation

The continuous wave Doppler system is the simplest and least expensive device
for measuring blood velocity. Two transducers are required, with one transmitting
the incident ultrasound and the other detecting the resultant continuous echoes
(Fig. 14-38). An oscillator produces a resonant frequency to drive the transmit transducer and provides the same frequency signal to the demodulator, which compares
the returning frequency to the incident frequency. The receiver amplifies the returning
signal and extracts the residual information containing the Doppler shift frequency
by using a low-pass filter, which removes the superimposed high-frequency oscillations. The Doppler signal contains very low-frequency signals from vessel walls
and other moving specular reflectors that a wall filter selectively removes. An audio
amplifier amplifies the Doppler signal to an audible sound level, and a recorder
tracks spectrum changes as a function of time for analysis of transient pulsatile flow.
Continuous wave Doppler suffers from depth selectivity with accuracy affected
by object motion within the beam path. Multiple overlying vessels will result in
superimposition, making it difficult to distinguish a specific Doppler signal. Spectral
Transmitter
amplifier
545
Oscillator
Transmitting
transducer
Receiver
amplifier
Receiving
transducer
Area of overlap and

location of vessel
Mixer
Demodulator
Wall Filter
Spectrum
analyzer
Headphones / speakers
Incident wave
Returning wave
Demodulator
Doppler Signal
FIGURE 14-38 Block diagram of a continuous wave Doppler system. Two transducers are required: one as a
transmitter and the other as a receiver. The area of overlap determines the position of blood velocity measurement. Signals from the receiver are mixed with the original frequency to extract the Doppler signal. A low-pass
filter removes the highest frequencies in the demodulated signals, and a high-pass filter (Wall filter) removes
the lowest frequencies due to tissue and transducer motion to extract the desired Doppler shift.
broadening of the frequencies occurs with a large sample area across the vessel profile
(composed of high velocity in the center and slower velocities at the edge of the vessels). Advantages of continuous mode include high accuracy of the Doppler shift
measurement because a narrow frequency bandwidth is used and no aliasing when
high velocities are measured, as occurs with pulsed Doppler operation (see below).
Quadrature Detection
The demodulation technique measures the magnitude of the Doppler shift but does
not reveal the direction of the Doppler shift, that is, whether the flow is toward or
away from the transducers. A method of signal processing called quadrature detection is phase sensitive and can indicate the direction of flow either toward or away
from the transducers.
Pulsed Doppler Operation

Pulsed Doppler ultrasound combines the velocity determination of continuous wave
Doppler systems and the range discrimination of pulse-echo imaging. A transducer tuned
for pulsed Doppler operation is used in a pulse-echo format, similar to imaging. The SPL
is longer (a minimum of 5 cycles per pulse up to 25 cycles per pulse) to provide a higher
Q factor and improve the measurement accuracy of the frequency shift (although at the
expense of axial resolution). Depth selection is achieved with an electronic time gate circuit to reject all echo signals except those falling within the gate window, as determined
by the operator. In some systems, multiple gates provide profile patterns of velocity
546
Transmit
Receive
Switch
Transmit
Gate
Clock
Oscillator
Pulsed
Doppler
transducer
Sample
range
Amplifier
De-modulator
Sample
length
Receive
Gate
Sample
volume
Length
delay
Range
delay
Sample
and
Hold
Headphones /
Speakers
Spectrum
Analyzer
Filter
FIGURE 14-39 Block diagram of a pulsed Doppler system. Isolation of a selected area is achieved by gating
the time of echo return and analyzing only those echoes that fall within the time window of the gate. In the
pulsed mode, the Doppler signal is discretely sampled in time to estimate the frequency shifts occurring in the
Doppler gate. Because axial resolution isnt as important as narrow bandwidths to better estimate the Doppler
shift, a long spatial pulse width (high Q factor) is employed.
values across a vessel. Figure 14-39 illustrates a simple block diagram of the pulsed
Doppler system and the system subcomponents necessary for data processing.
Each Doppler pulse does not contain enough information to completely determine the Doppler shift, but only a sample of the shifted frequencies measured as a
phase change. Stationary objects within the sample volume do not generate a phase
change in the returning echo when compared to the oscillator phase, but a moving
object does. Repeated echoes from the active gate are analyzed in the sample/hold
circuit, and a Doppler signal is gradually built up (Fig. 14-40A). The discrete measurements acquired at the PRF produce the synthesized Doppler signal. According to
sampling theory, a signal can be reconstructed unambiguously as long as the true frequency (e.g., the Doppler shift) is less than half the sampling rate. Thus, the PRF must
be at least twice the maximal Doppler frequency shift encountered in the measurement.
The maximum Doppler shift f max that is unambiguously determined in the
pulsed Doppler acquisition follows directly from the Doppler equation by substituting Vmax for V:
fmax =
PRF 2 f0 Vmax cos(q )

=
2
c
Rearranging the equation and solving for Vmax:

Vmax =
c PRF
4 f0 cos(q )
shows that the maximum blood velocity that is accurately determined is increased
with larger PRF, lower operating frequency, and larger angle (the cosine of the angle
gets smaller with larger angles from 0 to 90 degrees).
A
Pulse
547
Echo
Sample and Hold
Phase Change
Doppler
Signal
Wall Filter
Time
Sampling frequency
Low frequency
Doppler shift
> 2 samples/ cycle
High frequency
Doppler shift
< 2 samples/ cycle
Assigned (aliased) frequency
FIGURE 14-40 A. The returning ultrasound pulses from the Doppler gate are sampled over several pulseecho cycles (in this example, six times), in order to estimate Doppler shifts (if any) caused by moving blood cells.
A sample and hold circuit measures a variation of signal with time (note that the echo [purple line] and the
pulse [blue line] vary in phase, which are recorded and analyzed by Fourier transform methods to determine
the Doppler shift frequencies). The wall filter removes the low-frequency degradations caused by transducer
and patient motion. B. Aliasing occurs when the frequencies in the sampled signal are greater than the
PRF (sampling frequency). In this example, a signal of twice the frequency is analyzed as if it were the lower
frequency and thus mimics (aliases) the lower frequency.
For Doppler shift frequencies exceeding one-half the PRF, aliasing will occur,
causing a potentially significant error in the velocity estimation of the blood
(Fig. 14-40B). Thus, a 1.6-kHz Doppler shift requires a minimum PRF of 2 1.6 kHz =
3.2 kHz. One cannot simply increase the PRF to arbitrarily high values, because of echo
transit time and possible echo ambiguity. Use of a larger angle between the ultrasound
beam direction and the blood flow direction (e.g., 60 degrees) reduces the Doppler
shift. Thus, higher velocities can be unambiguously determined for a given PRF at larger
Doppler angles. However, at larger angles (e.g., 60 to 90 degrees), small errors in angle
estimation cause significant errors in the estimation of blood velocity (see Table 14-7).
A 180-degree phase shift in the Doppler frequency represents blood that is moving away from the transducer. Often, higher frequency signals in the spectrum will
be interpreted as lower frequency signals with a 180-degree phase shift, such that the
highest blood velocities in the center of a vessel are measured as having reverse flow.
This is another manifestation of aliasing.
548
Duplex Scanning
Duplex scanning refers to the combination of 2D B-mode imaging and pulsed Doppler
data acquisition. Without visual guidance to the vessel of interest, pulsed Doppler
systems would be of little use. A duplex scanner operates in the imaging mode and
creates a real-time image. The Doppler gate is positioned over the vessel of interest
with size (length and width) appropriate for evaluation of blood velocity, and at an
orientation (angle with respect to the interrogating US beam) that represents the
Doppler angle. When switched to Doppler mode, the scanner electronics determines
the proper timing to extract data only from within the user-defined gate.
Instrumentation for duplex scanning is available in several configurations. Most
often, electronic array transducers switch between a group of transducers used to create a B-mode image and one or more transducers used for the Doppler information.
The duplex system allows estimation of the blood velocity directly from the Doppler shift frequency, since the velocity of sound and the transducer frequency are
known, while the Doppler angle can be estimated from the B-mode image and input
into the scanner computer for calculation. Once the velocity is known, flow (in units
of cm3/s) is estimated as the product of the vessels cross-sectional area (cm2) times
the velocity (cm/s).
Errors in the flow volume may occur. The vessel axis might not lie totally within the
scanned plane, the vessel might be curved, or flow might be altered from the perceived
direction. The beam-vessel angle (Doppler angle) could be in error, which is much
more problematic for very large angles, particularly those greater than 60 degrees, as
explained previously. The Doppler gate (sample area) could be mispositioned or of
inappropriate size, such that the velocities are an overestimate (gate area too small) or
underestimate (gate area too large) of the average velocity. Noncircular cross sections
will cause errors in the area estimate, and therefore errors in the flow volume.
Multigate pulsed Doppler systems operate with several parallel channels closely
spaced across the lumen of a single large vessel. The outputs from all of the gates can
be combined to estimate the velocity profile across the vessel, which represents the
variation of flow velocity within the vessel lumen. Velocities mapped with a color
scale visually separate the flow information from the gray-scale image, and a realtime color flow Doppler ultrasound image indicates the direction of flow through
color coding. However, time is insufficient to complete the computations necessary
for determining the Doppler shifts from a large number of gates to get real-time
image update rates, particularly for those located at depth.
Color Flow Imaging

Color flow imaging provides a 2D visual display of moving blood in the vasculature,
superimposed upon the conventional gray-scale image as shown in Figure 14-41.
Velocities and directions are determined for multiple positions within a subarea of
the image and then color encoded (e.g., shades of red for blood moving toward the
transducer, and shades of blue for blood moving away from the transducer). Twodimensional color flow systems do not use the full Doppler shift information because
of a lack of time and/or a lack of parallel channels necessary for real-time imaging.
Instead, phase-shift autocorrelation or time domain correlation techniques are used.
Phase-shift autocorrelation is a technique to measure the similarity of one scan
line measurement to another when the maximum correlation (overlap) occurs. The
autocorrelation processor compares the entire echo pulse of one A-line with that
of a previous echo pulse separated by a time equal to the PRP. This self-scanning
algorithm detects changes in phase between two A-lines of data due to any Doppler
549
shift over the time t (Fig. 14-41A). The output correlation varies proportionately
with the phase change, which in turn varies proportionately with the velocity at the
point along the echo pulse trace. In addition, the direction of the moving object
(toward or away from the transducer) is preserved through phase detection of the
echo amplitudes. Generally, four to eight traces are used to determine the presence
of motion along one A-line of the scanned region. Therefore, the beam must remain
stationary for short periods of time before insonating another area in the imaging
volume. Additionally, because a gray-scale B-mode image must be acquired at the
Timing and Digital Logic
Transducer
array &
Beam
Former
Scan
Converter
Pulse-echo
Imaging
System
Image
Formatter
Color Display
Color-flow
active
area
Gray-scale
Active area
Autocorrelation
Detector
A-line #1(a)
DT
A-line #1(b)
Phase differences equivalent to Doppler shift;

Phase shift determined by correlation.
B
FIGURE 14-41 A. The color flow acquisition produces dynamic gray-scale B-mode images with colorencoded velocity maps in a user-defined active area of multiple gates. Inset. Autocorrelation detection
is a technique to rapidly determine phase changes (equivalent to Doppler shift) in areas of motion deduced
from two (or more) consecutive A-lines of data along the same direction. B. Patient data showing arterial and
venous flow, with color assignments (red towards and blue away from the transducer) depicting the direction
of blood flow. Within the image, the Doppler gate and Doppler angle are shown within the blue area, and the
bottom graph depicts the Spectral Doppler information extracted from the gate.
550
same time, the flow information must be interleaved with the image information.
The motion data are mapped with a color scale and superimposed on the gray-scale
image. FOV determines the processing time necessary to evaluate the color flow
data. A smaller FOV delivers a faster frame rate but, of course, sacrifices the area
evaluated for flow. One important consideration is keeping the beam at an angle to
the vessel axis to avoid a 90-degree angle. This can be achieved by electronic steering of the color flow ultrasound beams or by angling the array transducer relative to
the vessel axis (Fig. 14-41B).
Time domain correlation is an alternate method for color flow imaging. It is
based upon the measurement that a reflector has moved over a time t between
consecutive pulse-echo acquisitions (Fig. 14-42). Correlation mathematically determines the degree of similarity between two quantities. From echo train one, a series
of templates are formed, which are mathematically manipulated over echo train
two to determine the time shifts that result in the best correlation. Stationary reflectors need no time shift for high correlation; moving reflectors require a time t
(either positive or negative) to produce the maximal correlation. The displacement
of the reflector (x) is determined by the range equation as x = (c t)/2, where
c is the speed of sound. Measured velocity (Vm) is calculated as the displacement
divided by the time between pulses (the PRP): Vm = x/PRP. Finally, correction for
the angle () between the beam axis and the direction of motion (like the standard
Doppler correction) is V Vm/cos(). The velocity determines assignment of color,
and the images appear essentially the same as Doppler-processed images. Multiple
pulse-echo sequences are typically acquired to provide a good estimate of reflector displacements; frame update rates are reduced corresponding to the number
of repeated measurements per line. With time domain correlation methods, short
transmit pulses can be used, unlike the longer transmit pulses required for Doppler
acquisitions where longer pulses are necessary to achieve narrow bandwidth pulses.
This permits better axial resolution. Also, time domain correlation is less prone to
aliasing effects compared to Doppler methods because greater time shifts can be
tolerated in the returning echo signals from one pulse to the next, which means that
higher velocities can be measured.
There are several limitations with color flow imaging. Noise and clutter of slowly
moving, solid structures can overwhelm smaller echoes returning from moving blood
cells in color flow image. The spatial resolution of the color display is much poorer
than the gray-scale image, and variations in velocity are not well resolved in a large
vessel. Velocity calculation accuracy by the autocorrelation technique can be limited.
Since the color flow map does not fully describe the Doppler frequency spectrum,
FIGURE 14-42 Time domain correlation

uses a short SPL and an echo template
to determine positional change of moving reflectors from subsequent echoes.
The template scans the echo train to find
maximum correlation in each A-line; the
displacement between the maximum correlations of each A-line divided by the PRP
is the measured velocity.
Stationary
Echoes
Moving
Echoes
Correlation
Template
Maximum
cross-correlation
consecutive echoes
DT
Distance moved
corresponds to
velocity of objects
551
many color 2D units also provide a duplex scanning capability to provide a spectral
analysis of specific questionable areas indicated by the color flow examination. Aliasing artifacts due to insufficient sampling of the phase shifts are also a problem that
affects the color flow image, causing apparent reversed flow in areas of high velocity
(e.g., stenosis).
Doppler Spectral Interpretation

The Doppler signal is typically represented by a spectrum of frequencies resulting
from a range of velocities contained within the sampling gate at a specific point in
time. Blood flow can exhibit laminar, blunt, or turbulent flow patterns, depending upon the vessel wall characteristics, the size and shape of the vessel, and the
flow rate. Fast, laminar flow exists in the center of large, smooth wall vessels, while
slower blood flow occurs near the vessel walls, due to frictional forces. Turbulent
flow occurs at disruptions in the vessel wall caused by plaque buildup and stenosis.
A large Doppler gate that is positioned to encompass the entire lumen of the vessel
will contain a large range of blood velocities, while a smaller gate positioned in the
center of the vessel will have a smaller, faster range of velocities. A Doppler gate
positioned near a stenosis in the turbulent flow pattern will measure the largest
range of velocities.
With the pulsatile nature of blood flow, the spectral characteristics vary with
time. Interpretation of the frequency shifts and direction of blood flow is accomplished with the fast Fourier transform, which mathematically analyzes the detected
signals and generates amplitude versus frequency distribution profile known as the
Doppler spectrum. In a clinical instrument, the Doppler spectrum is continuously
updated in a real-time spectral Doppler display (Fig. 14-43). This information is displayed on the video monitor, typically below the 2D B-mode image, as a moving
trace, with the blood velocity (proportional to Doppler frequency) as the vertical axis
(from Vmax to Vmax) and time as the horizontal axis. The intensity of the Doppler
signal at a particular frequency and moment in time is displayed as the brightness at
that point on the display. Velocities in one direction are displayed as positive values
along the vertical axis and velocities in the other direction are displayed as negative
values. As new data arrive, the information is updated and scrolled from left to right.
Pulsatile blood takes on the appearance of a choppy sinusoidal wave through the
periodic cycle of the heartbeat.
Interpretation of the spectral display provides the ability to determine the presence of flow, the direction of flow, and characteristics of the flow. It is more difficult to determine a lack of flow, since it is also necessary to ensure that the lack
of signal is not due to other acoustical or electrical system parameters or problems.
The direction of flow (positive or negative Doppler shift) is best determined with a
small Doppler angle (about 30 degrees). Normal flow is typically characterized by a
specific spectral Doppler display waveform, which is a consequence of the hemodynamic features of particular vessels. Disturbed and turbulent flow produce Doppler
spectra that are correlated with disease processes. In these latter situations, the spectral curve is filled in with a wide distribution of frequencies representing a wide
range of velocities, as might occur with a vascular stenosis. Vascular impedance and
pulsatile velocity changes concurrent with the circulation can be tracked by Doppler spectrum techniques. Pertinent quantitative measures, such as pulsatility index:
PI = (max min)/average and resistive index: RI = (max min)/max, are dependent
on the characteristics of the Doppler spectral display (see Fig. 14-44 for description
of max, min, and average values extracted from the display).
552

Spectral Doppler Display
Frequency or velocity
Time
Doppler Spectrum
Amplitude
max
-fmax
(Max reverse)
Frequency
(Velocity)
fmax
(Max forward)
B
FIGURE 14-43 A. The spectral Doppler display is a plot of the Doppler shift frequency spectrum displayed
vertically, versus time, displayed horizontally. The amplitude of the shift frequency is encoded as gray-scale
or color intensity variations. Bottom graph: Two Doppler spectra are shown from the spectral display at two
discrete points in time, with amplitude (gray-scale variations) plotted versus frequency (velocity). A broad
spectrum (bandwidth) represents turbulent flow, while a narrow spectrum represents laminar flow within the
Doppler gate. B. Color flow image showing the active color area and the corresponding spectral Doppler display (below) determined from a gate volume positioned over a specific location in the vasculature. The resistive
index is calculated from the color intensity spectral display and indicated on the image as RI (note the electronic
calipers on the spectrum). The color scale values and velocities in this acquisition are calibrated in m/s.
0.9 vmax
0.5 vmax
553
max
Velocity
average
-0.5 vmax
-0.1 vmax
Aliased signals
min
Adjustment of spectral baseline
FIGURE 14-44 Left. Aliasing of the spectral Doppler display is characterized by wrap-around of the highest velocities to the opposite direction when the sampling (PRF) is inadequate. Right. Without changing the
overall velocity range, the spectral baseline is shifted to incorporate higher forward velocity and less reverse
velocity to avoid aliasing. The maximum, average, and minimum spectral Doppler display values allow quantitative determination of clinically relevant information such as pulsatility index and resistive index.
Velocity Aliasing
Aliasing, as described earlier, is an error caused by an insufficient sampling rate
(PRF) relative to the high-frequency Doppler signals generated by fast-moving
blood. A minimum of two samples per cycle of Doppler shift frequency is required
to unambiguously determine the corresponding velocity. In a spectral Doppler
display, the aliased signals wrap around to negative amplitude, masquerading as
reversed flow (Fig. 14-44, left). The most straightforward method to reduce or
eliminate the aliasing error is for the user to adjust the velocity scale to a wider
range, as most instruments have the PRF of the Doppler unit linked to the scale setting (a wide range delivers a high PRF). If the scale is already at the maximum PRF,
the spectral baseline, which represents 0 velocity (0 Doppler shift), can be readjusted
to allocate a greater sampling (frequency range) for reflectors moving toward the
transducer (Fig. 14-42, right). However, the minimum to the maximum Doppler
shift still cannot exceed PRF/2. In such a case, the baseline might be adjusted
to 0.1 Vmax to 0.9 Vmax, to allow most of the frequency sampling to be assigned
to the positive velocities (positive frequency shifts). From the adjusted spectral
Doppler display, the maximum, minimum, and average velocities of the blood flow
can be determined, and the resistive index as well as other pertinent vascular values
can be calculated.
Power Doppler
Doppler analysis places a constraint on the sensitivity to motion, because the signals
generated by motion must be extracted to determine velocity and direction from
the Doppler and phase shifts in the returning echoes within each gated region.
In color flow imaging, the frequency shift encodes the pixel value and assigns a
color, which is further divided into positive and negative directions. Power Doppler
is a signal processing method that relies on the total strength of the Doppler signal
(amplitude) and ignores directional (phase) information. The power (also known
as energy) mode of signal acquisition is dependent on the amplitude of all Doppler
554

Color Flow
Power
Red
Blue
Red
Monochrome
FIGURE 14-45 A comparison of color Doppler (left) and power Doppler (right) studies shows the enhanced
sensitivity of the power Doppler acquisition, particularly in areas perpendicular to the beam direction, where
the signal is lost in the color Doppler image. Flow directionality, however, is not available in the power Doppler image.
signals, regardless of the frequency shift. This dramatically improves the sensitivity to motion (e.g., slow blood flow) at the expense of directional and quantitative
flow information. Compared to conventional color flow imaging, power Doppler
produces images that have more sensitivity to motion and are not affected as much
by the Doppler angle (largely nondirectional). Aliasing is not a problem as only the
strength of the frequency shifted signals are analyzed, and not the phase. Greater
sensitivity allows detection and interpretation of very subtle and slow blood flow.
On the other hand, frame rates tend to be slower for the power Doppler imaging mode, and a significant amount of flash artifacts occur, which are related to
color signals arising from moving tissues, patient motion, or transducer motion. The
name power Doppler is sometimes mistakenly understood as implying the use of
increased transmit power to the patient, but, in fact, the power levels are typically
the same as in a standard color flow procedure. The difference is in the processing
of the returning signals, where sensitivity is achieved at the expense of direction and
quantitation. Images acquired with color flow and power Doppler are illustrated in
Figure 14-45.
14.8 Miscellaneous Ultrasound Capabilities
Distance, Area, and Volume Measurements

Measurements of distance, area, and volume are performed routinely in diagnostic ultrasound examinations. This is possible because the speed of sound in soft
tissue is known to within about 1% accuracy (1,540 15 m/s), and calibration of the instrument can be easily performed based on round-trip time of the
pulse and the echo. A notable example of common distance evaluations is for
fetal age determination by measuring fetal head diameter (Fig. 14-46). Measurement accuracy is achieved by careful selection of reference positions, such as the
leading edge to the leading edge of the reflector echo signals along the axis of
the beam, as these points are less affected by variations in echo signal amplitude.
Measurements between points along the direction of the ultrasound beam are usually more reliable, because of the better axial spatial resolution. Points measured
along the lateral plane have a tendency to be smeared out over a larger area due to
the poorer lateral resolution of the system. Thus, horizontal and oblique measurements will likely be less accurate. The circumference of a circular object can easily
A
Fetal gestational age (GA) measurements
BPD: Biparietal diameter

FL: Femur Length
555
HC: Head circumference

AC: Abdominal circumference
Structured report with composite data
FIGURE 14-46 A. Ultrasound provides accurate distance measurements. Fetal age is often determined by
biparietal diameter, circumference measurements (top), and femur length, abdominal circumference measurements (bottom). Based upon known correlation methods, the gestational age can be calculated for each of
the measurements. B. The structured report captures and summarizes the data in a tabular format as part of
the reporting mechanism.
be calculated from the measured diameter (d) or radius (r), using the relationship circumference = 2r = d. Distance measurements extend to two dimensions
(area) in a straightforward way by assuming a specific geometric shape. Similarly,
area measurements in a given image plane extend to 3D volumes by estimating the
slice thickness (elevational resolution).
Ultrasound Contrast Agents

Ultrasound contrast agents for vascular and perfusion imaging are becoming
extremely important from the clinical perspective. Most agents are comprised of
encapsulated microbubbles of 3 to 5 mm diameter containing air, nitrogen, or
insoluble gaseous compounds such as perfluorocarbons. Encapsulation materials,
such as human albumin, provide a container for the gas to maintain stability for
a reasonable time in the vasculature after injection. The natural tendency is for
the gas to diffuse rapidly into the bloodstream (e.g., within seconds) even when
556
encapsulated; successful contrast agents maintain stability over a period of time

that allows propagation to specific anatomical vascular areas that are targeted for
imaging. Because of the small size of the encapsulated bubbles, perfusion of tissues is also possible, but the bubbles must remain extremely stable during the time
required for tissue uptake.
The basis for generating an ultrasound signal is the large difference in acoustic
impedance between the gas and the fluids and tissues, as well as the compressibility
of the bubbles compared to the incompressible materials that are displaced. The
bubbles are small compared with the wavelength of the ultrasound beam and thus
become a point source of sound, producing reflections in all directions. In addition, the compressibility produces shifts in the returning frequency of the echoes,
called frequency harmonics (described in the next section), which are typically
higher than the original ultrasound frequency. To fully use the properties of contrast
agents, imaging techniques apart from standard B-mode scans are necessary and
are based upon the nonlinear compressibility of the gas bubbles and the frequency
harmonics that are generated (see below, e.g., pulse inversion harmonic imaging).
Destruction of the microbubbles occurs with the incident ultrasound pulse and
therefore requires temporal delays between images to allow circulating contrast
agent to appear.
Harmonic Imaging
Harmonic frequencies are integral multiples of the frequencies contained in an ultrasound pulse. A pulse with center frequency of f0 MHz, upon interaction with a
medium, will contain high-frequency harmonics of 2 f0, 3 f0, 4 f0, etc. These higher
frequencies arise through the vibration of encapsulated gas bubbles used as ultrasound contrast agents or with the nonlinear propagation of the ultrasound as it travels
through tissues. For contrast agents, the vibration modes of the encapsulated gas
reemit higher order harmonics due to the small size of the microbubbles (~3- to
6-m diameter) and the resultant contraction/expansion from the acoustic pressure variations. Harmonic imaging enhances contrast agent imaging by using a lowfrequency incident pulse and tuning the receiver (using a multifrequency transducer)
to higher frequency harmonics. This approach allows removal of echo clutter from
fundamental frequency reflections in the near field to improve the sensitivity to
the ultrasound contrast agent. Even though the returning harmonic signals have
higher attenuation (e.g., the first harmonic will have approximately twice the attenuation coefficient compared to the fundamental frequency), the echoes have to only
travel half the distance as the originating ultrasound pulse and thus have a relatively
large signal. For harmonic imaging, longer pulse lengths are often used to achieve
a higher transducer Q factor. This allows an easier separation of the frequency harmonics from the fundamental frequency. Although the longer SPL degrades axial
resolution, the benefits of harmonic imaging overcome the slight degradation in
axial resolution.
Based on the harmonic imaging work with microbubble contrast agents, native
tissue harmonic imaging (imaging higher frequency harmonics produced by tissues when using lower frequency incident sound waves) is now possible and in
common use. Harmonic frequencies are generated by the nonlinear distortion
of the wave as the high-pressure component (compression) travels faster than the
low-pressure component (rarefaction) of the acoustic wave in tissue. This wave
distortion, illustrated in Figure 14-47A, increases with depth and localizes in the
central area of the beam. The returning echoes comprising the harmonics travel
A
557
The ultrasound wave distorts as high pressure

travels faster than low pressure
Faster speed
Slower speed
Increasing depth of tissue
Area of higher order harmonics

Transmit at 2 MHz
Receive at 4 MHz
Harmonics are integer multiples of the transmit center frequency
Echo Intensity
Harmonics build in relative intensity with depth in

the tissues to a given depth
fo
Fundamental
2fo First
3fo Second
Depth
Receiver Bandwidth
Response
fo
2fo
3fo
fo
2fo
3fo
fo
2fo
3fo
Frequency spectrum of returning signal
FIGURE 14-47 A. Harmonic frequencies, integer multiples of the fundamental frequency, are produced
by the nonlinear propagation of the ultrasound beam, where the high-pressure component travels faster
than the low-pressure component of the wave. The wave distortion occurs in the central area of the beam.
B. Ultrasound harmonic frequencies (2f0, 3f0) build with depth and attenuate at a higher, frequency-dependent
rate. The frequency spectrum and harmonic amplitudes continuously change with depth (lower figure, displaying three points in time). The transducer bandwidth response must encompass the higher harmonics.
only slightly greater than one-half the distance to the transducer and, despite the
higher attenuation, have less but substantial amplitude compared to the fundamental frequency (Fig. 14-47B). The first harmonic (twice the fundamental frequency)
is commonly used because it suffers less attenuation than higher order harmonics
and because higher order harmonics are likely to exceed the transducers bandwidth.
558

C
Amplitude
Transducer Bandwidth Response

Transmit
Receive
fo
2fo
Frequency
FIGURE 14-47 (Continued) C. Native tissue harmonic imaging uses a lower center frequency spectrum
(e.g., 2-MHz spectrum) and receives echoes from the higher harmonics (e.g., 4-MHz spectrum).
Tuning the broadband receiver to the first harmonic spectrum filters out the lower
frequency echoes (to the extent that the spectra do not overlap) and eliminates ultrasound reflections and scattering from tissues and objects adjacent to the transducer
(Fig. 14-47C). Improved lateral spatial resolution (a majority of the echoes are produced in the central area of the beam), reduced side lobe artifacts, and removal of
multiple reverberation artifacts caused by anatomy adjacent to the transducer are
some advantages of tissue harmonic imaging. Comparison of conventional and harmonic right kidney images demonstrates improved image quality (Fig. 14-48), typical
of many examinations. While not always advantageous, native tissue harmonic imaging
Conventional
Harmonic
FIGURE 14-48 Conventional (left) and harmonic image (right) of the kidney demonstrate the improved
image quality of the harmonic image from reduced echo clutter, increased contrast, and resolution. (Images
provided by Dr. Kiran Jain, MD, University of California Davis Health System).
Curvilinear
array
Phased arrays
559
Intracavitary ultrasound probes
FIGURE 14-49 Transducer assemblies come in a variety of shapes and sizes. Each transducer (size, format,
frequency) is selected for specific examinations. External (left) and intracavitary (right) transducers are shown.
Intracavitary probes provide an inside out acoustic mapping of soft tissue anatomy.
is best applied in situations such as abdominal imaging that require a lower frequency
to achieve adequate penetration to begin with and then switched to the higher frequency harmonic for better quality and less clutter adjacent to the transducer.
Transducer Assemblies
Most general-purpose ultrasound units are equipped with at least a linear array
abdominal transducer of relatively low frequency (good penetration), and a separate small-parts phased-array transducer of high frequency (high spatial resolution).
Special-purpose transducers come in a wide variety of sizes, shapes, and operational
frequencies, usually designed for a specific application. Intracavitary transducers for
transrectal, transurethral, transvaginal, and transesophageal examinations provide
high-resolution, high-frequency imaging of anatomic structures that are in proximity
to the cavity walls. For example, transducers for transrectal scanning with linear array
or mechanical rotating designs are used routinely to image the bladder or the prostate
gland. External and intracavitary transducer probes are shown in Figure 14-49.
Intravascular ultrasound (IVUS) devices are catheter-mounted, typically with a
rotating single transducer or phased-array transducer (up to 64 acoustic elements)
design, and commonly used in interventional radiology and cardiology procedures.
Operating frequencies from about 10 MHz up to 60 MHz are typical, with PRF operation from 15 to 80 kHz create high-resolution (80 to 100 mm detail) acoustic images
inside the vessel. The high attenuation and minimal range of the ultrasound allow
high PRF sampling. A stack of cylindrical images can be created as the catheter is
pulled back through the vasculature of interest, and the vessel volume rendered.
Intravascular transducers can be used to assess vessel wall morphology, to differentiate plaque types (fibrous, fibrofatty, necrotic core, dense calcium), to estimate
stenosis, and to determine the efficacy of vascular intervention. Figure 14-50 shows
IVUS transducer types, depiction of the ultrasound beam, and cylindrical images
demonstrating a normal and a diseased vessel lumen.
Ultrasound Biopsy Guidance

Ultrasound is extremely useful for the guidance of biopsy procedures of suspicious
solid masses or signs of abnormal tissue change. A fine needle or a core needle is
used in many areas of the body, such as the breast, prostate, thyroid, abdominal and
560
FIGURE 14-50 Intravascular ultrasound: devices and images. Cathetermounted transducer arrays provide
an acoustic analysis of the vessel
lumen and wall from the inside out.
Mechanical (rotating shaft and acoustic mirror with a single transducer) and
electronic phased-array transducer
assemblies are used. Images show a
normal vessel lumen (lower left) and
reduced luminal stenosis and plaque
buildup (lower right).
Mechanical:
Transducer
Rotating acoustic beam
Electronic:
Radially directed US beam
Catheter
Multi-element transducer
Normal vasculature
Elliptical stenosis
pelvic areas. This is made possible with the excellent needle visibility due to the
reflection and scattering of ultrasound under real-time image acquisition.
Three-Dimensional Imaging
3D ultrasound imaging acquires 2D tomographic image data in a series of individual
B-scans of a volume of tissue. Forming the 3D dataset requires location of each individual 2D image using known acquisition geometries. Volume sampling can be achieved
in several ways with a transducer array: (1) linear translation, (2) freeform motion with
external localizers to a reference position, (3) rocking motion, and (4) rotation of the
scan (Fig. 14-51A). Three-dimensional image acquisition as a function of time (4D)
allows visualization of motion during the scan and rendering of the 3D data. With the
volume dataset acquisition geometry known, rendering into a surface display (Fig.
14-51B) by maximal intensity projection processing or volume surface rendering is
achieved with data reordering. Applications of various 3D imaging protocols are being
actively pursued, particularly in obstetric imaging. Features such as organ boundaries
are identified in each image, and the computer calculates the 3D surface, complete
with shading effects or false color for the delineation of anatomy.
14.9 Ultrasound Image Quality and Artifacts

Image quality is dependent on the design characteristics of the ultrasound equipment,
the numerous equipment variables selected, and positioning skills of the operator.
The equipment variables controlled by the operator include transducer frequency,
PRF, ultrasound intensity, and TGC curves, among others. Measures of ultrasound
image quality include spatial resolution, contrast resolution, image uniformity, and
noise characteristics. Image artifacts are common phenomena that can enhance or
degrade the diagnostic value of the ultrasound image.
Wedge
Linear
Freeform
561
Rotational
Surface-rendered 3D images
FIGURE 14-51 3D ultrasound acquisitions can be accomplished in several ways as depicted on the left in
the figure and include linear, wedge, freeform, and circular formats. Reconstruction of the dataset provides
3D surface-shaded and/or wire mesh renditions of the anatomy. The top images on the right are from a 4D
acquisition of a fetus at two different points in time. The bottom image shows the 3D surface evaluation of a
fetus with a cleft lip.
Spatial Resolution
Ultrasound spatial resolution has components in three directionsaxial, lateral,
and elevational. Axial resolution is determined by the frequency of the ultrasound
and the damping factor of the transducer, which together determine the spatial
pulse length. Lateral and elevational resolutions are determined by the dimensions
(width and height, respectively) of the transducer aperture, the depth of the object,
and mechanical and electronic focusing (see Figs. 14-19 to 14-23). Increasing the
depth of lateral resolution focus involves a trade-off of temporal resolution for realtime imaging. Axial and lateral resolutions are in the plane of the image and plainly
discernable, while elevational (slice-thickness) resolution is perpendicular to the
plane of the image and not as easy to understand or to interpret. Resolution in
the axial direction (along the beam) is equal to SPL and independent of depth.
Lateral and elevational resolutions are strongly dependent on depth. The minimum
resolution in the lateral/elevational directions is typically three to five times greater
than axial resolution.
Elevational resolution is a function of the height of the transducer array and
is depth dependent as dictated by the near field/far field beam characteristics of
the fixed transducer height. Poor elevational resolution occurs adjacent to the
transducer array and beyond the near/far field interface. Elevational focusing is
possible with an acoustic lens shaped along the height of the elements, which can
produce an elevational focal zone closer to the array surface. Alternatively, 1.5D
array transducers have several rows (typically five to seven) of independent elements in the elevational direction (1.5-D indicates that the number of elements
in the elevational direction is much less than in the lateral direction [over 100
562
to 200 elements]). Elevational focusing is achieved by introducing phase delays

among the elements in different rows to electronically focus the beam in the slicethickness dimension at a given depth (Fig. 14-23). Multiple elevational transmit
focal zones incur a time penalty similar to that required for the multiple lateral
focal zones.
Contrast Resolution and Noise

Contrast resolution depends on several interrelated factors. Acoustic impedance
differences (Table 14-3) give rise to reflections that delineate tissue boundaries and
internal architectures. The density and size of scatterers within tissues or organs
produce a specific texture (or lack of texture) that provides recognition and detection capabilities over the FOV. Ultrasound scattering introduces additive signals
from areas other than from echoes generated by specular reflectors and degrades
image contrast. With proper signal processing, attenuation differences (Table 14-5)
result in gray-scale differences among the tissues. Areas of low and high attenuation often produce distal signal enhancement or signal loss (e.g., fluid-filled cysts
[hypoechoic], gallstones) that allows detection and identification of tissues in the
image. Introduction of microbubble contrast agents improves the visualization of
the vasculature and tissue perfusion. Spatial compounding provides multiple beam
angles to better depict tissue boundaries, as well as provide averaging to reduce
stochastic speckle and electronic noise. Harmonic imaging improves image contrast
by eliminating unimportant or degrading signals from lower frequency echoes. In
addition, Doppler imaging techniques use moving anatomy and sophisticated processing techniques to generate contrast.
Contrast resolution also depends upon spatial resolution. Details within the image
are often distributed over the volume element represented in the tomographic slice
(e.g., Fig. 14-19), which varies in the lateral and elevational dimensions as a function
of depth. In areas where the slice thickness is relatively wide (close to the transducer
array surface and at great depth), the returning echoes generated by a small object are
averaged over the minimum volume, resulting in a lower signal and possible loss of
detection. On the other hand, objects larger than the minimum volume element can
actually achieve better contrast relative to the background because the random noise
components are reduced by averaging over the volume.
Detection of subtle anatomy in the patient is dependent on the contrast-tonoise ratio. The contrast is generated by differences in signal amplitude as discussed
above. Electronic noise is mainly generated by the electronic amplifiers of the system but is occasionally induced by environmental sources such as electrical power
fluctuations and equipment malfunction such as a dead or poorly functioning transducer element. A low-noise, high-gain amplifier is crucial for optimal low-contrast
resolution. Exponential attenuation of the ultrasound beam, which reduces contrast
and increases noise with depth, requires TGC to improve depth uniformity. Image
processing that specifically reduces noise, such as temporal or spatial averaging, can
increase the contrast-to-noise ratio; however, trade-offs include lower frame rates
and/or poorer spatial resolution. Low-power operation (e.g., an obstetrics power
setting with low transmit gain) requires higher electronic signal amplification to
increase the weak echo amplitudes to useful levels and results in a higher noise level
and lower contrast-to-noise ratio. Increasing the transmit power and/or the PRF can
improve contrast resolution, but there is a limit with respect to transducer capabilities, and, furthermore, the intensity must be restricted to levels unlikely to cause
biological damage.
563
Artifacts
Artifacts arise from the incorrect display of anatomy or noise during imaging. The
causes are machine and operator related, as well as intrinsic to the characteristics and
interaction of ultrasound with the tissues. Understanding how artifacts are generated
in ultrasound is crucial, which places high demands on the knowledge and experience of the sonographer.
Artifacts can be caused by a variety of mechanisms. For instance, sound travels
at different speeds, not just the 1,540 m/s average value for soft tissue that is used
in the range equation for placing echo information in the image matrix. This
speed variation results in some echoes being displaced from the expected location
in the image display. Sound is refracted when the beam is not perpendicular to
a tissue boundary; echoes are deflected to the receiver from areas outside of the
main beam and can be mismapped into the image. Improper use of TGC causes
suboptimal image quality with nonuniform appearance of tissues within a given
band of theimage.
Fortunately, most ultrasound artifacts are discernible to the experienced sonographer because of transient appearance and/or obvious effects on the image. Some artifacts are used to advantage as a diagnostic aid in characterization of tissue structures and
their composition. Besides aliasing in Doppler and color flow acquisitions discussed in
Section 14.7, many common artifacts in B-mode imaging are discussed below.
Refraction
Refraction is a change in the transmitted ultrasound pulse direction at a boundary
with nonperpendicular incidence, when the two tissues support a different speed
of sound (C1 C2). Misplaced anatomy often occurs in the image during the scan
(Fig.14-52A). The ultrasonographer must be aware of objects appearing and disappearing with slight differences in orientation of the beam. Anatomical displacement
due to refraction artifacts will change with the position of the transducer and angle
of incidence with the tissue boundaries.
Shadowing and Enhancement

Shadowing is a hypointense signal area distal to an object or interface and is
caused by objects with high attenuation or reflection of the incident beam without the return of echoes. Highly attenuating objects such as bones or kidney
stones reduce the intensity of the transmitted beam and can induce low-intensity
streaks in the image. Reflection of the incident beam from curved surfaces eliminates the distal propagation of ultrasound and causes streaks or shadowing.
Enhancement occurs distal to objects having very low ultrasound attenuation,
such as fluid-filled cavities (e.g., a filled bladder or cysts). Hyperintense signals
(through transmission) arise from increased transmission of sound by these
structures (Fig. 14-52B).
Reverberation
Reverberation artifacts arise from multiple echoes generated between two closely
spaced interfaces reflecting ultrasound energy back and forth during the acquisition
of the signal and before the next pulse. These artifacts are often caused by reflections
between a highly reflective interface and the transducer or between reflective interfaces such as metallic objects (e.g., bullet fragments), calcified tissues, or air pocket/
partial liquid areas of the anatomy. Reverberation echoes are typically manifested as
multiple, equally spaced boundaries with decreasing amplitude along a straight line
564
from the transducer (Fig. 14-52C). Comet tail artifact is a form of reverberation.
Ring-down artifacts arise from resonant vibrations within fluid trapped between a
tetrahedron of air bubbles, which creates a continuous sound wave that is transmitted back to the transducer and displayed as a series of parallel bands extending
posterior to a collection of gas.
Refraction
Image
C1
C1
C2
C2
B
Through
transmission
Reflection
Attenuation
Low
High
Through
transmission
Shadowing
FIGURE 14-52 A. Refraction is a change in the direction of the ultrasound beam that results from nonperpendicular incidence at a boundary where the two tissues do not have the same speed of sound. During a scan, anatomy can be missed and/or dislocated from the true position. B. Attenuation and reflection
of the ultrasound beam cause intensity changes. Enhancement occurs distal to objects of low attenuation,
manifested by a hyperintense signal, while shadowing occurs distal to objects of high attenuation, resulting
in a hypointense signal. At the curved edges of an organ boundary or mass, nonperpendicular reflection can
cause distal hypointense streaks and shadowing. Clinical images show through transmission (top) of a low
attenuation cyst and shadowing (bottom), caused by high attenuation gallstones. C. Reverberation commonly
occurs between to strong reflectors, such as an air pocket and the transducer array (left diagram) or with
calcium deposits and internal reflections (right image). The echoes bounce back and forth between the two
boundaries and produce equally spaced signals of diminishing amplitude in the image. This is often called a
comet-tail artifact.
565
Reverberation
Surface
Air
First echo
First reverb
Second reverb
Third reverb
Fourth reverb
Speed artifact
Liver
Fat
Boundary
displacement
E
Side Lobes
Image
pseudo
sludge
Alternate
view
FIGURE 14-52 (Continued) D. Speed of sound variation in the tissues can cause a mismapping of anatomy.
In the case of fatty tissues, the slower speed of sound in fat (1,450 m/s) results in a displacement of the returning echoes from distal anatomy by about 6% of the distance traveled through the mass. E. Side lobe energy
emissions in transducer arrays can cause anatomy outside of the main beam to be mapped into the main
beam. For a curved boundary, such as the gallbladder, side lobe interactions can be remapped and produce
findings such as pseudo sludge that is not apparent with other scanning angles. Clinical images (top) show
pseudosludge, which is not evident after repositioning the transducer assembly (bottom).
566

F
Color Doppler Mirror Image
Diaphragm
B-mode Mirror Image
Mirror
image
FIGURE 14-52 (Continued) F. A mirror image artifact arises from multiple beam reflections between a
mass and a strong reflector, such as the diaphragm. Multiple echoes result in the creation of a mirror image
of the mass beyond the diaphragm. Clinical images show mirror artifacts for color Doppler (top) and B-mode
scan (bottom), where the lower arrow points to the artifact mirrored by the diaphragm. G. Twinkling artifact
occurs with color flow imaging, caused by strong reflectors such as calcified stones, resulting in a changing
mixture of colors indicating flow. On the left is a B-mode image of the bladder and the corresponding color
flow image (inset trapezoidal area) demonstrating the twinkling artifact.
Speed Displacement
The speed displacement artifact is caused by the variability of speed of sound in
different tissues. In particular, the lower speed of sound in fat (1,450 m/s) causes
edge discontinuities of organ borders distal to fatty tissue (Fig. 14-52D). Edges are
mapped outward relative to the nondisplaced tissue borders by about 6% of the distance traveled through the fat (e.g., (1,540 1,450)/1,540 100 = 5.8%). Range and
distance uncertainty result from the speed artifact, and this also reduces the accuracy
of spatial measurements made with ultrasound.
Side Lobes and Grating Lobes

Side lobes are emissions of the ultrasound energy that occur in a direction slightly offaxis from the main beam (see Figs. 14-17 and 14-18) and arise from the expansion
of the piezoelectric elements orthogonal to the main beam. Echoes returning from
tissues along the propagation direction of the side lobes are positioned in the image
567
as if they occurred along the main beam. One type of side lobe artifact occurs near
a highly reflective surface just out of the main beam. Sometimes, side lobes redirect
diffuse echoes from adjacent soft tissues into an organ that is normally hypoechoic.
This occurs, for instance, in imaging of the gall bladder, where the side lobes produce
artifactual pseudosludge in an otherwise echo-free organ (Fig. 14-52E). Compared
to grating lobes, side lobes are more forward directed and are present with all types
of single-element and multielement transducer assemblies.
Grating lobes occur with multielement array transducers and result from the
division of a smooth transducer surface into a large number of small elements. Ultrasound energy is produced at a large angle relative to the main beam. This misdirected
energy can create ghost images of off-axis high-contrast objects. Grating lobe artifacts
are reduced by using very closely spaced elements in the array (less than one-half
wavelength apart). Linear array transducers are more prone to grating lobe artifacts
than phased-array transducers, chiefly due to the larger width and spacing of the
individual elements.
Multipath Reflection and Mirror Image

Near highly reflective surfaces, multiple beam reflections and refractions can find
their way back to the transducer. The anatomy involved in these reflections is misplaced on the beam axis more distal to the actual position caused by delays of the
echoes returning from the reflector(s). A common example is the interface of the liver
and the diaphragm in abdominal imaging. The pulse from the transducer generates
echoes from a mass in the liver and continues to the diaphragm, where a very strong
echo is produced. This echo travels from the diaphragm back to the mass, producing another set of echoes now directed back to the diaphragm. These echoes are rereflected from the diaphragm to the transducer. The back and forth travel distance of
the second echo set from the mass produces an artifact in the image that resembles a
mirror image of the mass, placed beyond the diaphragm (Fig. 14-52F).
Ambiguity
Ambiguity artifacts are created when a high PRF limits the amount of time spent
listening for echoes during the PRP. As the PRF increases, the PRP decreases, with
returning echoes still arriving from a greater depth after the next pulse is initiated.
Mismapping of very deep echoes to shallow positions can occur in the image.
Twinkling Artifact
Doppler mode detects motion, particularly blood flow, and displays moving blood as
red or blue (or selectable color) on the monitor, depending on direction and velocity. The twinkling artifact is represented as a rapidly changing mixture of colors, is
typically seen distal to a strong reflector such as a calculus, and is often mistaken for
an aneurysm when evaluating vessels. This artifactual appearance is possibly due to
echoes from the strong reflector with frequency changes due to the wide bandwidth
of the initial pulse and the narrow band ringing caused by the structure. Twinkling
artifact may be used to identify small renal stones, as shown in Figure 14-52G, and
differentiate echogenic foci from calcifications within kidney, gall bladder, and liver.
Slice Thickness
Slice thickness is determined by the beam width of the transducer array perpendicular to the image plane and is greater than the beam width in the image plane. The
thickness of the slice profile varies with depth, being broad close to the transducer,
narrowing at the elevational focal zone, and widening with distance beyond the focal
568
zone. Consequences of this slice-thickness shape are loss of signal from objects that
are much smaller than the volume element due to partial volume averaging and
inclusion of signals from highly reflective objects that are not in the imaging plane.
These artifacts are most significant at distances close and far from the transducer (see,
e.g., Fig. 14-53B).
14.10Ultrasound System Performance and Quality
Assurance
The system performance of a diagnostic ultrasound unit is described by several

parameters, including sensitivity and dynamic range, spatial resolution, contrast
sensitivity, range/distance accuracy, dead zone thickness, and TGC operation. For
Doppler studies, PRF, transducer angle estimates, and range gate stability are key
issues. To ensure the performance, accuracy, and safety of ultrasound equipment,
periodic QC measurements are recommended. The American College of Radiology
(ACR) has implemented an accreditation program that specifies recommended periodic QC procedures for ultrasound equipment. The periodic QC testing frequency
of ultrasound components should be adjusted to the probability of finding instabilities or maladjustment. This can be assessed by initially performing tests frequently,
reviewing logbooks over an extended period, and, with documented stability, reducing the testing rate.
Ultrasound Quality Control

Equipment QC is essentially performed every day during routine scanning by the
sonographer, who should and can recognize major problems with the images and the
equipment. Ensuring ultrasound image quality, however, requires implementation of
a QC program with periodic measurement of system performance to identify problems before serious malfunctions occur. Required are tissue-mimicking phantoms with
acoustic targets of various sizes and echogenic features embedded in a medium of uniform attenuation and speed of sound characteristic of soft tissues. Various multipurpose
phantoms are available to evaluate the clinical capabilities of the ultrasound system.
A generic phantom comprised of three modules is illustrated in Figure 14-53AC.
The phantom gel filler has tissue-like attenuation of 0.5 to 0.7 (dB/cm)/MHz (higher
attenuation provides a more challenging test) and low-contrast targets within a
matrix of small scatterers to mimic tissue background. Small, high-contrast reflectors
are positioned at known depths for measuring the axial and lateral spatial resolution,
for assessing the accuracy of horizontal and vertical distance measurements, and for
measuring the depth of the dead zone (the nonimaged area immediately adjacent to
the transducer). Another module contains low-contrast, small-diameter spheres (or
cylinders) of 2 and 4 mm diameter uniformly spaced with depth, to measure elevational resolution (slice-thickness) variation with depth. The third module is composed of a uniformly distributed scattering material for testing image uniformity and
penetration depth.
Spatial resolution, contrast resolution, and distance accuracy are evaluated with
one module (Fig. 14-53A). Axial resolution is evaluated by the ability to resolve highcontrast targets separated by 2, 1, 0.5, and 0.25 mm at three different depths. In an
optimally functioning system, the axial resolution should be consistent with depth
and improve with higher operational frequency. Lateral resolution is evaluated by
measuring the lateral spread of the high-contrast targets as a function of depth and
569
transmit focus. Contrast resolution is evaluated with gray-scale objects of lower

and higher attenuation than the tissue-mimicking gel; more sophisticated phantoms
have contrast resolution targets of varying contrast and size. Contrast resolution
should improve with increased transmit power. Dead zone depth is determined with
the first high-contrast target (positioned at several depths from 0 to ~1 cm) visible
in the image. Horizontal and vertical distance measurement accuracy uses the small
A
low scatter targets
dead zone targets
horizontal
accuracy
targets
Depth, cm
grayscale
targets
6
8
2
1
0.5
0.25
(mm)
10
12
14
axial
resolution
targets
16
18
vertical accuracy targets
Elevational resolution phantom

2 mm spheres
0.5 cm
tissue mimicking gel
Image of phantom
4 mm spheres
0.75 cm
Elevational focal
zone
Spheres positioned in the
center of the phantom
FIGURE 14-53 General-purpose ultrasound quality assurance phantoms are typically comprised of several
scanning modules. A. System resolution targets (axial and lateral), dead zone depth, vertical and horizontal
distance accuracy targets, contrast resolution (gray-scale targets), and low-scatter targets positioned at several depths (to determine penetration depths) are placed in a tissue-mimicking (acoustic scattering) gel. B.
Elevational resolution is determined with spheres equally distributed along a plane in tissue-mimicking gel. An
image of the phantom shows the effects of partial volume averaging and variations in elevational resolution
with depth. (Image reprinted by permission of Gammex, Inc., Madison, WI).
570

C
0
2
Vertical
band
Depth, cm
4
6
Horizontal
band
8
10
12
14
16
Maximum
Penetration
Depth
18
FIGURE 14-53 (Continued) C. The system uniformity module elucidates possible problems with image uniformity. Shown in this figure is a simulation of a transducer malfunction (vertical dropout) and horizontal transition mismatch that occurs with multiple lateral focusing zone image stitching. Penetration depth for uniform
response can also be determined.
igh-contrast targets. Vertical targets (along the axial beam direction) should have
h
higher precision and accuracy than the corresponding horizontal targets (lateral resolution), and all measurements should be within 5% of the known distance.
Elevational resolution and partial volume effects are evaluated with the sphere
module (Fig. 14-53B). The ultrasound image of the spherical targets illustrates the
effects of slice-thickness variation with depth and the dependence of resolvability on
object size. With the introduction of 1.5D transducer arrays as well as 3D imaging
capabilities, multiple transmit focal zones to reduce slice thickness at various depths
is becoming important, as is the need to verify elevational resolution performance.
Uniformity and penetration depth are measured with the uniformity module
(Fig. 14-53C). With a properly adjusted and operating ultrasound system, a uniform
response is expected up to the penetration depth capabilities of the transducer array.
Higher operational frequencies will display a lower penetration depth. Evidence
of vertical-directed shadowing for linear arrays or angular-directed shadowing for
phased arrays is an indication of malfunction of a particular transducer element or its
associated circuitry. Horizontal variations indicate inadequate handling of transitions
between focal zones when in the multifocal mode. These two problems are simulated
in the figure of the uniformity module.
During acceptance testing of a new unit, all transducers should be evaluated, and
baseline performance measured against manufacturer specifications. The maximum
depth of visualization is determined by identifying the deepest low-contrast scatterers
in a uniformity phantom that can be perceived in the image. This depth is dependent
on the type of transducer and its operating frequency and should be measured to
verify that the transducer and instrument components are operating at their designed
sensitivity levels. In a 0.7-dB/cm-MHz attenuation medium, a depth of 18 cm for
abdominal and 8 cm for small-parts transducers is the goal for visualization when
a single-transmit focal zone is placed as deeply as possible with maximum transmit
power level and optimal TGC adjustments. For a multifrequency transducer, the mid
frequency setting should be used. With the same settings, the uniformity section of
571
TABLE 14-8RECOMMENDED QC TESTS FOR ACR ACCREDITATION
PROGRAM
TEST (GRAY SCALE IMAGING MODE) FOR EACH SCANNER
MINIMUM FREQUENCY
System sensitivity and/or penetration capability
Semiannually
Image uniformity
Semiannually
Photography and other hard copy
Semiannually
Low contrast detectability (optional)
Semiannually
Assurance of electrical and mechanical safety
Semiannually
Horizontal and vertical distance accuracy
At acceptance
Transducers (of different scan format)
Ongoing basis
the phantom without targets assesses gray level and image uniformity. Power and
gain settings of the machine can have a significant effect on the apparent size of
point-like targets (e.g., high receive gain reveals only large-size targets and has poorer
resolution). One method to improve test reproducibility is to set the instrument to
the threshold detectability of the targets and to rescan with an increased transmit
gain of 20 dB.
Recommended QC procedures for ACR accreditation are listed in Table 14-8.
Routine QC testing must occur regularly. The same tests must be performed during
each testing period, so that changes can be monitored over time and effective corrective action can be taken. Testing results, corrective action, and the effects of corrective action must be documented and maintained on site. Other equipment-related
issues involve cleaning air filters, checking for loose or frayed cables, and checking
handles, wheels, and wheel locks as part of the QC tests.
The most frequently reported source of performance instability of an ultrasound
system is related to the display on maladjusted video monitors. Drift of the ultrasound instrument settings and/or poor viewing conditions (for instance, portable
ultrasound performed in a very bright patient room, potentially causing inappropriately gain-adjusted images) can lead to suboptimal images on the softcopy monitor.
The analog contrast and brightness settings for the monitor should be properly established during installation; monitor calibration should be performed according to the
DICOM Grayscale Standard Display Function at least semiannually (see Chapter 5)
and verified with image test patterns (e.g., the SMPTE pattern).
Doppler Performance Measurements

Doppler techniques are becoming more common in the day-to-day use of medical
ultrasound equipment. Reliance on flow measurements to make diagnoses requires
demonstration of accurate data acquisition and processing. QC phantoms to assess
velocity and flow contain one or more tubes in tissue-mimicking materials at various
depths. A blood-mimicking fluid is pushed through the tubes with carefully calibrated pumps to provide a known velocity for assessing the accuracy of the Doppler
velocity measurement. Several tests can be performed, including maximum penetration depth at which flow waveforms can be detected, alignment of the sample volume with the duplex B-mode image, accuracy of velocity measurements, and volume
flow. For color-flow systems, sensitivity and alignment of the color flow image with
the B-scan gray-scale image are assessed.
572
14.11 Acoustic Power and Bioeffects

Power is the rate of energy production, absorption, or flow. The SI unit of power is
the watt (W), defined as one joule of energy per second. Acoustic intensity is the rate
at which sound energy flows through a unit area and is usually expressed in units of
watts per square centimeter (W/cm2) or milliwatts per square centimeter (mW/cm2).
Ultrasound acoustic power levels are strongly dependent on the operational characteristics of the system, including the transmit power, PRF, transducer frequency,
and operation mode. Biological effects (bioeffects) are predominately related to the
heating of tissues caused by high intensity levels of ultrasound used to enhance image
quality and functionality. For diagnostic imaging, the intensity levels are kept below
the threshold for documented bioeffects.
Acoustic Power and Intensity of Pulsed Ultrasound

Measurement Methods
Measurement of ultrasound pressure amplitude within a beam is performed with a
hydrophone, a device containing a small (e.g., 0.5-mm-diameter) piezoelectric element coupled to external conductors and mounted in a protective housing. When
placed in an ultrasound beam, the hydrophone produces a voltage that is proportional to the variations in pressure amplitude at that point in the beam as a function
of time, permitting determination of peak compression and rarefaction amplitude
as well as pulse duration and PRP (Fig. 14-54A). Calibrated hydrophones provide
absolute measures of pressure, from which the acoustic intensity can be calculated if
the acoustic impedance of the medium is accurately known.
Intensity Measures of Pulsed Ultrasound

In the pulsed mode of ultrasound operation, the instantaneous intensity varies greatly
with time and position. At a particular location in tissue, the instantaneous intensity
is quite large while the ultrasound pulse passes through the tissue, but the pulse
duration is only about a microsecond or less, and for the remainder of the PRP, the
intensity is nearly zero.
The temporal peak, ITP, is the highest instantaneous intensity in the beam, the temporal average, ITA, is the time-averaged intensity over the PRP, and the pulse average,
IPA, is the average intensity of the pulse (Fig. 14-54B). The spatial peak, ISP, is the highest intensity spatially in the beam, and the spatial average, ISA, is the average intensity
over the beam area, usually taken to be the area of the transducer (Fig. 14-54C).
The acoustic power contained in the ultrasound beam (watts), averaged over at
least one PRP and divided by the beam area (usually the area of the transducer face),
is the spatial averagetemporal average intensity ISATA. Other meaningful measures for
pulsed ultrasound intensity are determined from ISATA, including
1. The spatial averagepulse average intensity, ISAPA = ISATA/duty cycle, where IPA = ITA/
duty cycle
2. The spatial peaktemporal average intensity, ISPTA = ISATA [ISP/ISA], which is a good
indicator of thermal ultrasound effects
3. The spatial peakpulse average intensity, ISPPA = ISATA [ISP/ISA]/duty cycle, an indicator of potential mechanical bioeffects and cavitation
For acoustic ultrasound intensity levels, ISPPA > ISPTA > ISAPA > ISATA. Typical acoustical power
outputs are listed in Table 14-9. The two most relevant measures are the ISPPA and the ISPTA.
Pressure Amplitude
573
+
Peak Compression
Hydrophone
measurement
0
Peak Rarefaction
Time
Intensity
Temporal Peak (TP)
Pulse Average (PA)

Temporal Average (TA)
0
Time
C
Intensity
Spatial Peak (SP)
Beam
Intensity
Profile
Spatial Average (SA)
Distance across beam

FIGURE 14-54 A. Pressure amplitude variations are measured with a hydrophone and include peak compression and peak rarefaction variations with time. B. Temporal intensity variations of pulsed ultrasound vary
widely, from the temporal peak and temporal average values; pulse average intensity represents the average
intensity measured over the pulse duration. C. Spatial intensity variations of pulsed ultrasound are described
by the spatial peak value and the spatial average value, measured over the beam profile.
Both measurements are required by the US Food and Drug Administration (FDA) for
certification of instrumentation. Values of ISPTA for diagnostic imaging are usually below
100 mW/cm2 for imaging, but for certain Doppler applications, ISPTA can exceed 1,000
mW/cm2. ISPPA can be several orders of magnitude greater than ISPTA, as shown in
Table 14.9. For real-time scanners, the combined intensity descriptors must be modified to consider dwell time (the time the ultrasound beam is directed at a particular region) and the acquisition geometry and spatial sampling. These variations help
explain the measured differences between the ISPTA and ISPPA values indicated, which are
much less than the duty cycle values that are predicted by the equations above.
TABLE 14-9TYPICAL INTENSITY MEASURES FOR ULTRASOUND DATA
COLLECTION MODES
MODE
PRESSURE
AMPLITUDE (MPa)
ISPTA (mW/cm2)
ISPPA (W/cm2)
POWER (mW)
B-scan
1.68
19
174
18
M-mode
1.68
73
174
Pulsed Doppler
2.48
1,140
288
31
Color flow
2.59
234
325
81
Adapted from compilation of data presented by the American Institute of Ultrasound in Medicine. Note the
difference in units for ISPTA (mW/cm2) versus ISPPA (W/cm2).
574
Thermal and mechanical indices of ultrasound operation are now the accepted
method of determining power levels for real-time instruments that provide the operator with quantitative estimates of power deposition in the patient. These indices are
selected for their relevance to risks from biological effects and are displayed on the
monitor during real-time scanning. The sonographer can use these indices to minimize power deposition to the patient (and fetus) consistent with obtaining useful clinical images in the spirit of the ALARA (As Low As Reasonably Achievable) concept.
Thermal Index
The thermal index, TI, is the ratio of the acoustical power produced by the transducer
to the power required to raise tissue in the beam area by 1C. This is estimated by the
ultrasound system using algorithms that take into account the ultrasonic frequency,
beam area, and the acoustic output power of the transducer. Assumptions are made
for attenuation and thermal properties of the tissues with long, steady exposure
times. An indicated TI value of 2 signifies a possible 2C increase in the temperature
of the tissues when the transducer is stationary. TI values are associated with the ISPTA
measure of intensity.
On some scanners, other thermal indices that might be encountered are TIS (S
for soft tissue), TIB (B for bone), and TIC (C for cranial bone). These quantities are
useful because of the increased heat buildup that can occur at a bonesoft tissue
interface when present in the beam, particularly for obstetric scanning of late-term
pregnancies, and with the use of Doppler ultrasound (where power levels can be
substantially higher).
Mechanical Index
Cavitation is a consequence of the negative pressures (rarefaction of the mechanical
wave) that induce bubble formation from the extraction of dissolved gases in the
medium. The mechanical index, MI is a value that estimates the likelihood of cavitation by the ultrasound beam. The MI is directly proportional to the peak rarefactional
(negative) pressure and inversely proportional to the square root of the ultrasound
frequency (in MHz). An attenuation of 0.3 (dB/cm)/MHz is assumed for the algorithm
that estimates the MI. As the ultrasound output power (transmit pulse amplitude)
is increased, the MI increases linearly, while an increase in the transducer frequency
(say from 2 to 8 MHz) decreases the MI by the square root of 4 or by a factor of two.
MI values are associated with the ISPPA measure of intensity.
Biological Mechanisms and Effects

Diagnostic ultrasound has established a remarkable safety record. Significant deleterious bioeffects on either patients or operators of diagnostic ultrasound imaging procedures have not been reported in the literature. Despite the lack of evidence that any
harm can be caused by diagnostic intensities of ultrasound, it is prudent and indeed
an obligation of the physician to consider issues of benefit versus risk when performing an ultrasound exam and to take all precautions to ensure maximal benefit with
minimal risk. The American Institute of Ultrasound in Medicine recommends adherence to the ALARA principles. US FDA requirements for new ultrasound equipment
include the display of acoustic output indices (MI and TI) to give the user feedback
regarding the power deposition to the patient.
At high intensities, ultrasound can cause biological effects by thermal and mechanical mechanisms. Biological tissues absorb ultrasound energy, which is converted
575
into heat; thus, heat will be generated at all parts of the ultrasonic field in the tissue.
Thermal effects are dependent not only on the rate of heat deposition in a particular
volume of the body but also on how fast the heat is removed by blood flow and other
means of heat conduction. The best indicator of heat deposition is the ISPTA measure
of intensity and the calculated TI value. Heat deposition is determined by the average ultrasound intensity in the focal zone and the absorption coefficient of the tissue.
Absorption increases with the frequency of the ultrasound and varies with tissue
type. Bone has a much higher attenuation (absorption) coefficient than soft tissue,
which can cause significant heat deposition at a tissue-bone interface. In diagnostic
ultrasound applications, the heating effect is typically well below a temperature rise
(e.g., 1C to 2C) that would be considered potentially damaging, although some
Doppler instruments can approach these levels with high pulse repetition frequencies and longer pulse duration.
Nonthermal mechanisms include mechanical movement of the particles of the
medium due to radiation pressure (which can cause force or torque on tissue structures)
and acoustic streaming, which can give rise to a steady circulatory flow. With higher
energy deposition over a short period, cavitation can occur, broadly defined as sonically
generated activity of highly compressible bodies composed of gas and/or vapor. Cavitation can be subtle or readily observable and is typically unpredictable and sometimes
violent. Stable cavitation generally refers to the pulsation (expansion and contraction) of
persistent bubbles in the tissue that occur at low and intermediate ultrasound intensities (as used clinically). Chiefly related to the peak rarefactional pressure, the MI is an
estimate for producing cavitation. At higher ultrasound intensity levels, transient cavitation can occur, whereby the bubbles respond nonlinearly to the driving force, causing
a collapse approaching the speed of sound. At this point, the bubbles might dissolve,
disintegrate, or rebound. In the minimum volume state, conditions exist that can dissociate the water vapor into free radicals such as H and OH, which can cause chemical
damage to biologically important molecules such as DNA. Short, high amplitude pulses
such as those used in imaging are good candidates for transient cavitation; however, the
intensities used in diagnostic imaging are far below the transient cavitation threshold
(e.g., 1 kW/cm2 peak pulse power is necessary for transient cavitation to be evoked).
Although biological effects have been demonstrated at much higher ultrasound
power levels and longer durations, the levels and durations for typical imaging and
Doppler studies are below the threshold for known undesirable effects. At higher
output power, outcomes include macroscopic damage (e.g., rupturing of blood
vessels, breaking up cellsindeed, the whole point of shock wave lithotripsy
the breakup of kidney stones) and microscopic damage (e.g., breaking of chromosomes, changes in cell mitotic index). No bioeffects have been shown below
ISPTA of 100 mW/cm2 (Fig. 14-55). Even though ultrasound is considered safe when
used properly, prudence dictates that ultrasound exposure be limited to only those
patients for whom a definite benefit will be obtained.
14.12 Summary
Ultrasound uses mechanical energy to generate acoustic maps of the body, which
can be extremely useful in providing valuable diagnostic information that can be
acquired in a very safe and efficient manner. Quality of the exam is extremely operator dependent, and of all medical imaging modalities, ultrasound can be considered an art and the operator must have extensive knowledge and understanding
of the underlying basic physics to be considered a true artist. This is also true of
576

100
Intensity (W/cm2)
10
"Potentially harmful zone"
"Safe zone"
0.1
Diagnostic Ultrasound levels
0.01
1
10
100
Exposure time, minutes
1000
10000
FIGURE 14-55 A diagram of potential bioeffects from ultrasound delineates safe and potentially harmful
regions according to ultrasound intensity levels and exposure time. The dashed line shows the upper limit of
intensities typically encountered in diagnostic imaging applications.
the radiologist who must use this physics knowledge to understand the ultrasound
interactions, generation of the image, limitations, potential artifacts, and possible
pitfalls that are present in order to make a confident diagnosis. Despite the purported
safety of ultrasound, precautions must be taken to ensure the appropriateness of the
examination and the potential misdiagnoses and deleterious bioeffects that can occur
with misuse.
REFERENCES AND Suggested Readings

American Institute of Ultrasound in Medicine, Official Statements and Reports, available at the AIUM web
site: http://www.aium.org
Feldman MK. US Artifacts. Radiographics 29:11791189, 2009.
Hangiandreou N. AAPM/RSNA Physics Tutorial for Residents: Topics in US B-mode US: Basic Concepts
and New Technology. Radiographics 23:10191033, 2003.
Hedrick WR, Hykes DL, Starchman DE. Ultrasound Physics and Instrumentation, Fourth Edition.
St. Louis, MO: Mosby, 2005.
Kremkau FW. Sonography, principles and instruments. 8th ed. St. Louis, MO: Elsevier Saunders, 2011.
Zagzebski JA. Essentials of ultrasound physics. St. Louis, MO: Mosby, Inc., 1996.
Section
III
Nuclear Medicine
Chapter
15
Radioactivity and Nuclear

Transformation
15.1 Radionuclide Decay Terms and Relationships
Activity
The quantity of radioactive material, expressed as the number of radioactive atoms
undergoing nuclear transformation per unit time (t), is called activity (A). Described
mathematically, activity is equal to the change (dN) in the total number of radioactive
atoms (N) in a given period of time (dt), or

A = dN /dt
[15-1]
The minus sign indicates that the number of radioactive atoms decreases with time.
Activity has traditionally been expressed in units of curies (Ci). The curie is named
after Marie Curie, a Polish-born French chemist and physicist, who won two Nobel
prizes for her work with radioactive materials. One for physics in 1903 (with her
husband Pierre Curie and Henri Becquerel) for research on radioactivity and another
for chemistry in 1911 for her discovery of the elements radium and polonium, the
latter named after her native Poland. The term radioactivity was coined by Marie
Curie reflecting the intense radiation emissions from radium. One Ci is defined as
3.70 1010 disintegrations per second (dps), which is roughly equal to the rate of
disintegrations from 1 g of radium-226 (Ra-226). A curie is a large amount of radioactivity. In nuclear medicine, activities from 0.1 to 30 mCi of a variety of radionuclides are typically used for imaging studies, and up to 300 mCi of iodine-131 is used
for therapy. Although the curie is still the most common unit of radioactivity in the
United States, the majority of the worlds scientific literature uses the Systeme International (SI) units. The SI unit for radioactivity is the becquerel (Bq), named after
Henri Becquerel, who discovered radioactivity in 1896. The becquerel is defined
as 1 dps. One millicurie (mCi) is equal to 37 megabecquerels (1 mCi = 37 MBq).
Table 15-1 lists the units and prefixes describing various amounts of r adioactivity.
Decay Constant
Radioactive decay is a random process. From moment to moment, it is not possible
to predict which radioactive atoms in a sample will decay. However, observation of
a larger number of radioactive atoms over a period of time allows the average rate
of nuclear transformation (decay) to be established. The number of atoms decaying
per unit time (dN/dt) is proportional to the number of unstable atoms (N) that are
present at any given time:

dN/ dt N
[15-2]
579
580
Section III Nuclear Medicine
TABLE 15-1UNITS AND PREFIXES ASSOCIATED WITH VARIOUS QUANTITIES
OF RADIOACTIVITY
QUANTITY
SYMBOL
DPS
DPM
Gigabecquerel
GBq
1 109
6 1010
Megabecquerel
MBq
1 106
6 107
Kilobecquerel
kBq
1 10
6 104
Curie
Ci
Millicurie
mCi (10
Microcurie
3.7 1010
2.22 1012
3.7 10
2.22 109
Ci (106 Ci)
3.7 104
2.22 106
Nanocurie
nCi (10
3.7 10
2.22 103
Picocurie
pCi (1012 Ci)
3.7 102
2.22
Ci)
Ci)
Multiply mCi by 37 to obtain MBq or divide MBq by 37 to obtain mCi (e.g., 1 mCi = 37 MBq).
A proportionality can be transformed into an equality by introducing a constant. This

constant is called the decay constant ().

dN/ dt = N
[15-3]
The minus sign indicates that the number of radioactive atoms decaying per unit time
(the decay rate or activity of the sample) decreases with time. The decay constant is
equal to the fraction of the number of radioactive atoms remaining in a sample that
decay per unit time. The relationship between activity and l can be seen by considering Equation 15-1 and substituting A for dN/ dt in Equation 15-3:
A = lN
[15-4]
The decay constant is characteristic of each radionuclide. For example, the decay constants for technetium-99m (Tc-99m) and molybdenum-99 (Mo-99) are 0.1151 h1
and 0.252 day1, respectively.
Physical Half-Life
A useful parameter related to the decay constant is the physical half-life (T or
Tp). The half-life is defined as the time required for the number of radioactive
atoms in a sample to decrease by one half. The number of radioactive atoms remaining in a sample and the number of elapsed half-lives are related by the following
equation:

N = N 0 /2n
[15-5]
where N is number of radioactive atoms remaining, N0 is the initial number of radioactive atoms, and n is the number of half-lives that have elapsed. The relationship
between time and the number of radioactive atoms remaining in a sample is demonstrated with Tc-99m (Tp 6 h) in Table 15-2.
After 10 half-lives, the number of radioactive atoms in a sample is reduced by
approximately a factor of a thousand. After 20 half-lives, the number of radioactive
atoms is reduced to approximately one millionth of the initial number.
The decay constant and the physical half-life are related as follows:

l = ln2 / Tp = 0.693 / Tp
[15-6]
Chapter 15 Radioactivity and Nuclear Transformation
581
TABLE 15-2 RADIOACTIVE DECAYa

TIME (D)
NO. OF PHYSICAL
HALF-LIVES
EXPRESSION
N0/2
10
0.25
N0/2
5 10
0.5
N0/22
2.5 105
0.75
2.5
5
0
1
(N/N0) 100 % REMAINING
100%
50%
25%
N0/2
1.25 10
12.5%
N0/24
6.25 104
6.25%
10
N0/2
10
0.1% (1/1,000)N0
20
N0/220
10
0.000001% (1/1,000,000)N0
The influence of radioactive decay on the number of radioactive atoms in a sample is illustrated with
technetium-99m, which has a physical half-life of 6 h (0.25 days). The sample initially contains one million (106)
radioactive atoms (N).
a
where ln 2 denotes the natural logarithm of 2. Note that the derivation of this relationship is identical to that between the half value layer (HVL) and the linear attenuation coefficient () in Chapter 3 (Equation 3-9).
The physical half-life and the decay constant are physical quantities that are
inversely related and unique for each radionuclide. Half-lives of radioactive materials
range from billions of years to a fraction of a second. Radionuclides used in nuclear
medicine typically have half-lives on the order of hours or days. Examples of Tp and
for radionuclides commonly used in nuclear medicine are listed in Table 15-3.
TABLE 15-3PHYSICAL HALF-LIFE (Tp) AND DECAY
CONSTANT () FOR RADIONUCLIDES

USED IN NUCLEAR MEDICINE
RADIONUCLIDE
Tp
Rubidium-82 ( Rb)
75 s
0.0092 s1
Fluorine-18 (18F)
110 min
0.0063 min1
6.02 h
0.1151 h1
13.27 h
0.0522 h1
Samarium-153 ( Sm)
1.93 d
0.3591 d1
Yttrium-90 (90Y)
2.69 d
0.2575 d1
Molybdenum-99 ( Mo)
2.75 d
0.2522 d1
Indium-111 (111In)
2.81 d
0.2466 d1
Thallium-201 ( Tl)
3.04 d
0.2281 d1
Gallium-67 (67Ga)
3.26 d
0.2126 d1
Xenon-133 ( Xe)
5.24 d
0.1323 d1
Iodine-131 (131I)
8.02 d
0.0864 d1
Phosphorus-32 ( P)
14.26 d
0.0486 d1
Strontium-82 (82Sr)
25.60 d
0.0271 d1
Chromium-51 ( Cr)
27.70 d
0.0250 d1
Strontium-89 (89Sr)
50.53 d
0.0137 d1
Iodine-125 ( I)
59.41 d
0.0117 d1
Cobalt-57 (57Co)
271.79 d
0.0025 d1
82
Technetium-99m (
Tc)
99m
Iodine-123 (123I)
153
99
201
133
32
51
125
582
Fundamental Decay Equation

By applying the integral calculus to Equation 15-3, a useful relationship is established between the number of radioactive atoms remaining in a sample and time
the fundamental decay equation:

N t = N 0e t or At = A0e t
[15-7]
where Nt is the number of radioactive atoms at time t, At is the activity at time t, N0 is

the initial number of radioactive atoms, A0 is the initial activity, e is the base of natural
logarithm 2.718, is the decay constant ln 2/Tp 0.693/Tp, and t is the
time.
Problem: A nuclear medicine technologist injects a patient with 400 Ci of indium111labeled autologous leukocytes (Tp 2.81 days). Twenty-four hours later, the
patient is imaged. Assuming that none of the activity was excreted, how much activity remains at the time of imaging?
Solution:
A A0et
Given:
A0 400 Ci
0.693/2.81 days 0.247 days1
t 1 day
Note: t and Tp must be in the same units of time.
1
At 400 Ci e(0.247 days ) (1 day)
At 400 Ci e0.247
At (400 Ci)(0.781)
At 312 Ci
A plot of activity as a function of time on a linear scale results in a curvilinear
exponential relationship in which the total activity asymptotically approaches zero
(Fig.15-1). If the logarithm of the activity is plotted versus time (semilog plot), this
exponential relationship appears as a straight line (Fig. 15-2).
15.2 Nuclear Transformation

As mentioned previously, when an unstable (i.e., radioactive) atomic nucleus undergoes the spontaneous transformation, called radioactive decay, radiation is emitted. If the
daughter nucleus is stable, this spontaneous transformation ends. If the daughter is also
100
Percent of Initial Activity
(Linear Scale)
FIGURE 15-1 Percentage of initial

activity as a function of time (linear
scale).
80
60
50
40
25
12.5
0
2
3
Time in Half-Lives
FIGURE 15-2 Percentage of initial

activity as a function of time (semilog
plot).
100
Percent of Initial Activity
(Log Scale)
583
50
40
30
25
20
12.5
10
2
Time in Half-Lives
unstable, the process continues until a stable nuclide is reached. Most radionuclides
decay in one or more of the following ways: (1) alpha decay, (2) beta-minus emission,
(3) beta-plus (positron) emission, (4) electron capture, or (5) isomeric transition.
Alpha Decay
Alpha () decay is the spontaneous emission of an alpha particle (identical to a
helium nucleus consisting of two protons and two neutrons) from the nucleus
(Fig.15-3). Alpha decay typically occurs with heavy nuclides (A . 150) and is often
followed by gamma and characteristic x-ray emission. These photon emissions are
often accompanied by the competing processes of internal conversion and Auger
electron emission. Alpha particles are the heaviest and least penetrating form of
radiation considered in this chapter. They are emitted from the atomic nucleus with
discrete energies in the range of 2 to 10 MeV. An alpha particle is approximately four
times heavier than a proton or neutron and carries an electronic charge twice that of
the proton. Alpha decay can be described by the following equation:
A
Z
A-4
Z-2
Y + 42 He 2 + + transition energy
alpha
particle
[15-8]
Example:
220
86
Rn
216
84
Po + 42He2+ + 6.4 MeV transition energy
Alpha decay results in a large energy transition and a slight increase in the ratio of neutrons to protons (N/Z ratio):
220
86
Rn
N/Z 134/86 1.56
216
84
Po
N/Z 132/84 1.57
+2
Alpha particle (He2+) decay
FIGURE 15-3 Alpha decay.
584
Alpha particles are not used in medical imaging because their ranges are limited to
approximately 1 cm/MeV in air and typically less than 100 m in tissue. Even the
most energetic alpha particles cannot penetrate the dead layer of the skin. However,
the intense ionization tracks produced by this high LET radiation (e.g., mean LET of
alpha particles is ~100 keV/m compared to ~3 keV/m for energetic electrons set
in motion by the interaction of diagnostic x-rays in tissue) make them a potentially
serious health hazard should alpha-emitting radionuclides enter the body via ingestion, inhalation, or a wound. Research continues to assess the potential therapeutic
effectiveness of alpha-emitting radionuclides such as astatine-212, bismuth-212, and
bismuth-213 (At-211, Bi-212, and Bi-213) chelated to monoclonal antibodies to
produce stable radioimmunoconjugates directed against various tumors as radioimmunotherapeutic agents.
Beta-Minus (Negatron) Decay

Beta-minus () decay, or negatron decay, characteristically occurs with radionuclides that have an excess number of neutrons compared with the number of
protons (i.e., a high N/Z ratio). Beta-minus decay can be described by the following equation:

A
Z
A
Z+1Y
energy

(negatron) (antineutrino)
[15-9]
This mode of decay results in the conversion of a neutron into a proton with the simultaneous ejection of a negatively charged beta particle () and an antineutrino ( v ),
(Fig. 15-4). With the exception of their origin (the nucleus), beta particles are identical to ordinary electrons. The antineutrino is an electrically neutral subatomic
particle whose mass is much smaller than that of an electron. The absence of charge
and the infinitesimal mass of antineutrinos make them very difficult to detect
because they rarely interact with matter. Beta decay increases the number of protons by 1 and thus transforms the atom into a different element with an atomic
number Z 1. However, the concomitant decrease in the neutron number means
that the mass number remains unchanged. Decay modes in which the mass number
remains constant are called isobaric transitions. Radionuclides produced by nuclear
fission are neutron rich, and therefore most decay by emission. Beta-minus
decay decreases the N/Z ratio, bringing the daughter closer to the line of stability
(see Chapter 2):
FIGURE 15-4 Beta-minus decay.
Beta Minus Decay
+
v
585
Example:
32
15
P
N / Z = 17 / 15 = 1.13
32
16
S
N / Z = 16 / 16 = 1.00
Although the particles emitted by a particular radionuclide have a discrete maximal energy (Emax), almost all are emitted with energies lower than the maximum. The
average energy of the particles is approximately 1/3 Emax. The balance of the energy
is given to the antineutrino (i.e., E max E - E ) . Thus, beta-minus decay results
in a polyenergetic spectrum of energies ranging from zero to Emax (Fig. 15-5). Any
excess energy in the nucleus after beta decay is emitted as gamma rays, internal conversion electrons, and other associated radiations.
Beta-Plus Decay (Positron Emission)

Just as beta-minus decay is driven by the nuclear instability caused by excess neutrons, neutron-poor radionuclides (i.e., those with a low N/Z ratio) are also unstable. Many of these radionuclides decay by beta-plus (positron) emission, which
increases the neutron number by one. Beta-plus decay can be described by the
following equation:
A
Z
energy
A
Z1
(positron)
(neutrino)
[15-10]
The net result is the conversion of a proton into a neutron with the simultaneous ejection of the positron () and a neutrino (). Positron decay decreases the number of
protons (atomic number) by 1 and thereby transforms the atom into a different element
with an atomic number of Z1 (Fig. 15-6). The daughter atom, with one less proton
in the nucleus, initially has one too many orbital electrons and thus is a negative ion.
However, the daughter quickly releases the extra orbital electron to the surrounding
FIGURE 15-5 P-32 example. Distribution of beta-minus particle kinetic

energy. Number of beta-minus particles emitted per 106 decays of P-32
as a function of energy.
P-32 Decay: Beta Energy Spectrum

1,000
Betas per 106
decays per keV
800
I
n
t
e
n
s
i
t
y
Eavg=695 keV
600
Emax=1,711 keV
400
200
500
1,000
Beta Energy (keV)
1,500
2,000
586

FIGURE 15-6 Beta-plus decay.
medium and becomes a neutral atom. Position decay can only occur if the mass of
the parent atom exceeds that of the daughter atom by at least the masses of the two
electrons (positron and orbital electron). According to Einsteins mass-energy equivalence formula, E mc2, 511 keV is the energy equivalent of the rest mass of an electron (positively or negatively charged). Therefore, there is an inherent threshold for
positron decay equal to the sum of the rest mass energy equivalent of two electrons
(i.e., 2 511 keV, or 1.02 MeV).
The number of neutrons is increased by 1; therefore, the transformation is isobaric because the total number of nucleons is unchanged. Accelerator-produced radionuclides, which are typically neutron deficient, often decay by positron emission.
Positron decay increases the N/Z ratio, resulting in a daughter closer to the line of
stability.
Example:
18
9
F
N / Z 9 / 9 1
18
8
O
N / Z 10 / 8 1.25
The energy distribution between the positron and the neutrino is similar to that
between the negatron and the antineutrino in beta-minus decay; thus positrons are
polyenergetic with an average energy equal to approximately 1/3 Emax. As with
decay, excess energy following positron decay is released as gamma rays and other
associated radiation.
Although decay has similarities to decay, there are also important
differences. The neutrino and antineutrino are antiparticles, as are the positron
and negatron. The prefix anti- before the name of an elementary particle denotes
another particle with certain symmetry characteristics. In the case of charged particles such as the positron, the antiparticle (i.e., the negatron) has a charge equal but
opposite to that of the positron and a magnetic moment that is oppositely directed
with respect to spin. In the case of neutral particles such as the neutrino and antineutrino, there is no charge; therefore, differentiation between the particles is made
solely on the basis of differences in magnetic moment. Other important differences
between the particle and antiparticle are their lifetimes and their eventual fates. As
mentioned earlier, negatrons are physically identical to ordinary electrons and as
such lose their kinetic energy as they traverse matter via excitation and ionization.
When they lose all (or almost all) of their kinetic energy, they may be captured
587
by an atom or absorbed into the free electron pool. Positrons undergo a similar
process of energy deposition via excitation and ionization; however, when they
come to rest, they react violently with their antiparticles (electrons). This process
results in the entire rest mass of both particles being instantaneously converted to
energy and emitted as two oppositely directed (i.e., ~180 degrees apart) 511-keV
annihilation photons (Fig. 15-7). Medical imaging of annihilation radiation from positron-emitting radiopharmaceuticals, called positron emission tomography (PET),
is discussed in Chapter 19.
Electron Capture Decay

Electron capture () is an alternative to positron decay for neutron-deficient radionuclides. In this decay mode, the nucleus captures an orbital (usually a K- or L-shell)
electron, with the conversion of a proton into a neutron and the simultaneous ejection of a neutrino (Fig. 15-8). Electron capture can be described by the following
equation:
A
Z
X e-
A
Z1
(neutrino)
energy .
[15-11]
The net effect of electron capture is the same as positron emission: the atomic number is decreased by 1, creating a different element, and the mass number remains
unchanged. Therefore, electron capture is isobaric and results in an increase in the
N/Z ratio.
Example:
201
81
Tl
N / Z 120 / 81 1.48
201
80
Hg
N / Z 121/ 80 1.51
energy
The capture of an orbital electron creates a vacancy in the electron shell, which
is filled by an electron from a higher energy shell. As discussed in Chapter 2, this
electron transition results in the emission of characteristic x-rays and/or Auger
electrons. For example, thallium-201, (Tl-201) decays to mercury-201 (Hg-201)
by electron capture, resulting in the emission of characteristic x-rays. It is these
0.511 MeV
Annihilation Radiation
~180
+
Excitation and Ionization
0.511 MeV
FIGURE 15-7 Annihilation radiation.
588
FIGURE 15-8 Electron capture decay.
x-rays that are primarily used to create the images in Tl-201 myocardial perfusion
studies. As with other modes of decay, if the nucleus is left in an excited state
following electron capture, the excess energy will be emitted as gamma rays and
other radiations.
As previously mentioned, positron emission requires a mass energy difference
between the parent and daughter atoms of at least 1.02 MeV. Neutron-poor radionuclides below this threshold energy decay exclusively by electron capture. Nuclides
for which the energy difference between the parent and daughter exceed 1.02 MeV
may decay by electron capture or positron emission, or both. Heavier proton-rich
nuclides are more likely to decay by electron capture, whereas lighter proton-rich
nuclides are more likely to decay by positron emission. This is a result of the closer
proximity of the K- or L-shell electrons to the nucleus and the greater magnitude
of the coulombic attraction from the positive charges. Although capture of a K- or
L-shell electron is the most probable, electron capture can occur with higher energy
shell electrons.
The quantum mechanical description of the atom is essential for understanding electron capture. The Bohr model describes electrons in fixed orbits at discrete
distances from the nucleus. This model does not permit electrons to be close
enough to the nucleus to be captured. However, the quantum mechanical model
describes orbital electron locations as probability density functions in which
there is a finite probability that an electron will pass close to or even through the
nucleus.
Electron capture radionuclides used in medical imaging decay to atoms in excited
states that subsequently emit externally detectable x-rays, gamma rays, or both.
Isomeric Transition
Often during radioactive decay, a daughter is formed in an excited (i.e., unstable)
state. Gamma rays are emitted as the daughter nucleus undergoes an internal rearrangement and transitions from the excited state to a lower energy state. Once created, most excited states transition almost instantaneously (on the order of 1012 s)
to lower energy states with the emission of gamma radiation. However, some excited
states persist for longer periods, with half-lives ranging from nanoseconds (109 s)
to more than 30 years. These excited states are called metastable or isomeric states
and those with half-lives exceeding a millisecond (103 s) are denoted by the letter
m after the mass number (e.g., Tc-99m). Isomeric transition is a decay process that
yields gamma radiation without the emission or capture of a particle by the nucleus.
There is no change in atomic number, mass number, or neutron number. Thus, this
589
decay mode is isobaric and isotonic, and it occurs between two nuclear energy states
with no change in the N/Z ratio.
Isomeric transition can be described by the following equation:
Am
Z
X AZ X (energy)
[15-12]
The energy is released in the form of gamma rays, internal conversion electrons, or both.
Decay Schemes
Each radionuclides decay process is a unique characteristic of that radionuclide.
The majority of the pertinent information about the decay process and its associated
radiation can be summarized in a line diagram called a decay scheme (Fig. 15-9).
Decay schemes identify the parent, daughter, mode of decay, energy levels including
those of excited and metastable states, radiation emissions, and sometimes physical half-life and other characteristics of the decay sequence. The top horizontal line
represents the parent, and the bottom horizontal line represents the daughter. Horizontal lines between those two represent intermediate excited or metastable states.
By convention, a diagonal arrow to the right indicates an increase in Z, which occurs
with beta-minus decay. A diagonal arrow to the left indicates a decrease in Z such as
decay by electron capture. A vertical line followed by a diagonal arrow to the left is
used to indicate alpha decay and in some cases to indicate positron emission when
a radionuclide decays by both electron capture and positron emission (e.g., F-18).
Vertical down pointing arrows indicate gamma ray emission, including those emitted during isomeric transition. These diagrams are often accompanied by decay data
tables, which provide information on all the significant ionizing radiations emitted
from the atom as a result of the nuclear transformation. Examples of these decay
schemes and data tables are presented in this section.
Figure 15-10 shows the alpha decay scheme of radon-220 (Rn-220). Rn-220
has a physical half-life of 55 s and decays by one of two possible alpha transitions.
Alpha 1 (1) at 5.747 MeV occurs 0.07% of the time and is followed immediately by
a 0.55-MeV gamma ray (1) to the ground state. The emission of alpha 2 (2) with an
energy of 6.287 MeV, occurs 99.9% of the time and leads directly to the ground state.
The decay data table lists these radiations together with the daughter atom, which
has a 2 charge and a small amount of kinetic energy as a result of recoil from the
alpha particle emission.
Phosphorus-32 (P-32) is used in nuclear medicine as a therapeutic agent in the
treatment of a several diseases, including polycythemia vera, and serous effusions.
P-32 has a half-life of 14.3 days and decays directly to its ground state by emitting
a beta-minus particle with an Emax of 1.71 MeV (Fig. 15-11). The average (mean)
energy of the beta-minus particle is approximately 1/3 Emax (0.6948 MeV), with the
Electron capture decay

+
decay
AY
z1
A
X
z
E Higher
n
e
r Lower
g
y
N
u
c
l
e
a S Ground
r t
a
t
e
Parent
decay
AY*
z+1
AY
z+1
Excited state (*)

Isomeric transition
(Gamma Rays)
Daughter
FIGURE 15-9 Elements of the generalized decay scheme.
590
FIGURE 15-10 Principal decay scheme of

radon-220.
RADON-220
Alpha Decay
T1/2 = 55 seconds
220
Rn
86
0.5500 MeV
0.0 MeV
6.3992 MeV
1
216
Po
84
Decay Data Table
Radiation
Mean
Mean Energy
Number per
per Particle
Disintegration
(MeV)
Alpha
1
Recoil Atom
Alpha
2
Recoil Atom
Gamma
1
0.0007
0.0007
0.9993
0.9993
0.0006
5.7470
0.1064
6.2870
0.1164
0.5500
antineutrino carrying off the balance of the transition energy. There are no excited
energy states or other radiation emitted during this decay; therefore, P-32 is referred
to as a pure beta emitter.
A somewhat more complicated decay scheme is associated with the beta-minus
decay of Mo-99 to Tc-99 (Fig. 15-12). Eight of the ten possible beta-minus decay
transitions are shown with probabilities ranging from 0.822 for beta-minus 8 (i.e.,
82.2% of all decays of Mo-99 are by 8 transition) to 0.0004 (0.04%) for 6. The
sum of all transition probabilities (1 to 10) is equal to 1. The average energy of
beta particles from the transition is 0.4519 MeV. The 8 transition leads directly to a
metastable state of technetium 99, Tc-99m, which is 0.1427 MeV above the ground
state and decays with a half-life of 6.02 h. Tc-99m is the most widely used radionuclide in nuclear medicine.
After beta decay, there are a number of excited states created that transition
to lower energy levels via the emission of gamma rays and/or internal conversion
FIGURE 15-11 Principal decay scheme of phosphorus-32.
PHOSPHORUS-32
Beta-Minus Decay
T1/2 = 14.3 days
32
P
15
1.71 MeV
0.0 MeV
32
S
16
Stable
Decay Data Table
Radiation
Mean
Number per
Disintegration
Mean Energy
per Particle
(MeV)
Beta Minus
1.000
0.6948
591
MOLYBDENUM-99
Beta-Minus Decay
T1/2 = 2.75 days
99
Mo
42
1.1421
2
3
4
1.1297
1.0042
0.9208
6 5
0.6715
0.5090
7
12 1110 9
0.1810
0.1427
3 1 0.1405
2
0.0
2.12 105Y
8
99
Tc
43
6.02h Tc-99m
Decay Data Table

Mean
Mean Energy
Number per per Particle
Radiation
Disintegration
(MeV)
Radiation
Beta Minus
Beta Minus
Beta Minus
Beta Minus
Beta Minus
Beta Minus
Gamma
K Int Con Elect
L Int Con Elect
M Int Con Elect
Gamma
K Int Con Elect
L Int Con Elect
Gamma
K Int Con Elect
L Int Con Elect
M Int Con Elect
1
3
4
6
7
8
1
0.0010
0.0014
0.1640
0.0004
0.0114
0.8220
0.0105
0.0428
0.0053
0.0017
0.0452
0.0058
0.0007
0.0600
0.0085
0.0012
0.0004
0.0658
0.1112
0.1331
0.2541
0.2897
0.4428
0.0406
0.0195
0.0377
0.0401
0.1405
0.1194
0.1377
0.1811
0.1600
0.1782
0.1806
Gamma
Gamma
Gamma
Gamma
Gamma
Gamma
K Int Con Elect
Gamma
K Int Con Elect
Gamma
Gamma
K Alpha-1 X-Ray
K Alpha-2 X-Ray
K Beta-1 X-Ray
KLL Auger Elect
KLX Auger Elect
LMM Auger Elect
MXY Auger Elect
Mean
Mean Energy
Number per per Particle
Disintegration
(MeV)
4
5
6
7
8
9
10
11
12
0.0119
0.0001
0.0002
0.0006
0.0002
0.1367
0.0002
0.0426
0.0000
0.0013
0.0010
0.0253
0.0127
0.0060
0.0087
0.0032
0.0615
0.1403
0.3664
0.4706
0.4115
0.5288
0.6207
0.7397
0.7186
0.7779
0.7571
0.8230
0.9608
0.0183
0.0182
0.0206
0.0154
0.0178
0.0019
0.0004
FIGURE 15-12 Principal decay scheme of molybdenum-99. Auger electron nomenclature: KXY Auger Elect is
an auger electron emitted from the Y shell as a result of a transition of an electron of the X shell to a vacancy
in the K shell. X and Y are shells higher than the K shell. For example, KLL Auger Elect is an auger electron
emitted from the L shell as a result of a transition of another L shell electron to a vacancy in the K shell.
e lectrons. As previously described, the ejection of an electron by internal conversion

of the gamma ray results in the emission of characteristic x-rays, Auger electrons, or
both. All of these radiations, their mean energies, and their associated probabilities
are included in the decay data table.
The process of gamma ray emission by isomeric transition is of primary importance to nuclear medicine, because most procedures performed depend on the emission and detection of gamma radiation. Figure 15-13 shows the decay scheme for
Tc-99m. There are three gamma ray transitions as Tc-99m decays to Tc-99. The
gamma 1 (1) transition occurs very infrequently because 99.2% of the time this
energy is internally converted resulting in the emission of either an M-shell internal
conversion electron (86.2%) with a mean energy of 1.8 keV or an N-shell internal
conversion electron (13.0%) with a mean energy of 2.2 keV. After internal conversion, the nucleus is left in an excited state, which is followed almost instantaneously
by gamma 2 (2) transition at 140.5 keV to ground state. The 2 transition occurs
89.1% of the time with the balance of the transitions from 140.5 keV to ground state
592
TECHNETIUM 99m
Isomeric Transition
T1/2 = 6.02 hrs.
99m
Tc
43
0.1427
1
0.1405
0.0
99
2.12x105Y
Tc
43
Decay Data Table

Mean
Number per
Disintegration
Radiation
Gamma
M Int Con Elect
N Int Con Elect
Gamma
K Int Con Elect
L Int Con Elect
M Int Con Elect
Gamma
K Int Con Elect
L Int Con Elect
M Int Con Elect
K Alpha-1 X-Ray
K Alpha-2 X-Ray
K Beta-1 X-Ray
KLL Auger Elect
KLX Auger Elect
LMM Auger Elect
MXY Auger Elect
Mean Energy
per Particle
(MeV)
0.0000
0.8620
0.1300
0.8910
0.0892
0.0109
0.0020
0.0003
0.0088
0.0035
0.0011
0.0441
0.0221
0.0105
0.0152
0.0055
0.1093
1.2359
0.0021
0.0018
0.0022
0.1405
0.1194
0.1375
0.1377
0.1426
0.1215
0.1398
0.1422
0.0183
0.0182
0.0206
0.0154
0.0178
0.0019
0.0004
FIGURE 15-13 Principal decay scheme of technetium-99m. Auger electron nomenclature: KXY Auger Elect is
an auger electron emitted from the Y shell as a result of a transition of an electron of the X shell to a vacancy
in the K shell. X and Y are shells higher than the K shell. For example, KLL Auger Elect is an auger electron
emitted from the L shell as a result of a transition of another L shell electron to a vacancy in the K shell.
occurring primarily via internal conversion. Gamma 2 is the principal photon imaged
in nuclear medicine. Like gamma 1, the gamma 3 transition at 142.7 keV occurs very
infrequently relative to the probability of internal conversion electron emission. Here
again, the vacancies created in orbital electron shells following internal conversion
result in the production of characteristic x-rays and Auger electrons.
FIGURE 15-14 Principal decay
scheme of fluorine-18.
FLUORINE-18
Electron Capture and Beta-Plus Decay
T1/2 = 110 minutes
18
F
9
EC1
1.02 MeV
+
0.0 MeV
Stable
18
O
8
Decay Data Table
Radiation
Beta Plus
Annih. Radiation
Mean
Number per
Disintegration
0.9700
1.9400
1.655 MeV
Mean Energy
Particle
(MeV)
0.2496
0.5110
0.635 MeV
593
TABLE 15-4SUMMARY OF RADIONUCLIDE DECAY
TYPE OF
DECAY
Alpha
Beta
minus
PRIMARY
OTHER
RADIATION RADIATION
EMITTED
EMITTED
NUCLEAR TRANSFORMATION
He+2 ( )
4
2
-rays
C x-rays
AE, ICE
A
Z
Z > 83
N/Z too large
X ZA24 X 2 + 24 He2 + + energy 2
-rays
C x-rays
NUCLEAR
CONDITION
CHANGE IN
PRIOR TO
Z
A
TRANSFORMATION
A
Z
X ZA+1Y + + + energy
A
Z
X ZA1Y + + + + energy
N/Z too small
A
Z
X + e ZA1Y + + energy
N/Z too small
Excited or metastable nucleus
AE, ICE,
Beta
plus
-rays
C x-rays
AE, ICE, n
Electron C x-rays
capture
-rays
AE,ICE, n
Isomeric -rays
transition
C x-rays
Am
Z
X ZA X + energy
AE, ICE
ICE, internal conversion e1; AE, auger e1; C x-rays, Characteristic x-rays; -rays, Gamma rays;
n , antineutrino;
n, neutrino.
As discussed previously, positron emission and electron capture are competing decay processes for neutron-deficient radionuclides. As shown in Figure 15-14,
fluorine-18 (F-18) decays by both modes. F-18 decays by positron emission (represented by a solid vertical line followed by a diagonal arrow to the left) 97% of the
time. The length of the vertical part of the line in the diagram represents the sum of
the rest mass energy equivalent of the positron and electron (i.e., 1.02 MeV). Electron
capture (represented by a diagonal arrow to the left) occurs 3% of the time. The dual
mode of decay results in an effective decay constant (e) that is the sum of the positron (1) and electron capture (2) decay constants: e 1 2. The decay data table
shows that positrons are emitted 97% of the time with an average energy of 0.2496
MeV (~1/3 of 0.635 MeV, which is Emax). Furthermore, the interaction of the positron
with an electron results in the production of two 511-keV annihilation radiation
photons. Because two photons are produced for each positron, their abundance is
2 97%, or 194%. 18F is the most widely used radionuclide for PET imaging.
A summary of the characteristics of radionuclide decay modes previously discussed is provided in table 15.4
SUGGESTED READINGS
Cherry SR, et al. Physics in nuclear medicine. 4th ed. Philadelphia, PA: Saunders, 2011.
Friedlander G, Kennedy JW, Miller JM. Nuclear and radiochemistry. 3rd ed. New York, NY: Wiley, 1981.
Patton JA. Introduction to nuclear physics. Radiographics 1998;18:9951007.
Chapter
16
Radionuclide Production,
Radiopharmaceuticals, and
Internal Dosimetry
16.1 Radionuclide Production
Although many naturally occurring radioactive nuclides exist, all of those commonly
administered to patients in nuclear medicine are artificially produced. Artificial
radioactivity was discovered in 1934 by Irene Curie (daughter of Marie and Pierre
Curie) and Frederic Joliot, who induced radioactivity in boron and aluminum targets
by bombarding them with alpha () particles from polonium. Positrons continued
to be emitted from the targets after the alpha source was removed. Today, more than
2,500 artificial radionuclides have been produced by a variety of methods. Most
radionuclides used in nuclear medicine are produced by particle accelerators (e.g.,
cyclotrons), nuclear reactors, or radionuclide generators.
Cyclotron-Produced Radionuclides
Cyclotrons and other charged-particle accelerators produce radionuclides by bombarding stable nuclei with high-energy charged particles. Positively charged ions
such as protons (H), deuterons (2H), and alpha particles (4He2) as well as negatively charged hydrogen ions (H) are commonly used to produce radionuclides
used in medicine. Charged particles must be accelerated to high kinetic energies to
overcome and penetrate the repulsive coulombic barrier of the target atoms nuclei.
In 1930, Cockcroft and Walton applied a clever scheme of cascading a series
of transformers, each capable of stepping up the voltage by several hundred thousand volts. The large potentials generated were used to produce artificial radioactivity by accelerating particles to high energies and bombarding stable nuclei
with them.
In Berkeley, California, in 1931, E.O. Lawrence capitalized on this development
but added a unique dimension in his design of the cyclotron (Fig. 16-1). A cyclotron
has a vacuum chamber between the poles of an electromagnet. Inside the vacuum
chamber is a pair of hollow, semicircular electrodes, each shaped like the letter D
and referred to as dees. The two dees are separated by a small gap. An alternating
high voltage is applied between the two dees. When positive ions are injected into
the center of the cyclotron they are attracted to and accelerated toward the negatively
charged dee. The static magnetic field constrains the ions to travel in a circular path,
whereby the radius of the circle increases as the ions gain kinetic energy (Fig. 16-2).
Half way around the circle, the ions approach the gap between the dees; at this time,
the polarity of the electrical field between the two dees is reversed, causing the ions
to be accelerated toward the negative dee. This cycle is repeated again and again,
594
Chapter 16Radionuclide Production, Radiopharmaceuticals, and Internal Dosimetry 595

Magnetic field
Deflector (D)
Electrode
Dee(vacuum)
Alternating
Voltage
Exit window (W)
+/
To Target
Dee(vacuum)
Ion Source (S)
Top and bottom magnet removed
Magnetic field
North Pole
Vacuum
Dee
Gap
South Pole
Side View
FIGURE 16-1 Schematic view of a cyclotron. Two "dees" (A and B) are seperated by a small gap.
with the particles accelerated each time they cross the gap, acquiring kinetic energy
and sweeping out larger and larger circles. As the length of the path between successive accelerations increases, the speed of the particle also increases; hence, the time
interval between accelerations remains constant. The cyclic nature of these events led
FIGURE 16-2 A constant magnetic field imposes a force (F) on a
moving charged particle that is perpendicular to the direction of the
particles velocity (v). This causes an
ion in a cyclotron to move in a circular path. The diameter of the circular
path is proportional to the speed of
the ion.
Force on Particle from

Static Magnetic Field(F)
Direction of
Particle Motion(v)
F
v
596
to the name cyclotron. The final kinetic energy achieved by the accelerated particles depends on the type of particle (e.g., protons or deuterons), diameter of the dees,
and the strength of the static magnetic field. Finally, as the ions reach the periphery of
the dees, they are removed from their circular path by a negatively charged deflector
plate (if positive ions are accelerated) or electron stripping foil (if H ions are accelerated), emerge through the window, and strike the target. Depending on the design of
the cyclotron, particle energies can range from a few million electron volts (MeV) to
several hundred MeV.
The accelerated ions collide with the target nuclei, causing nuclear reactions.
An incident particle may leave the target nucleus after interacting, transferring
some of its energy to it, or it may be completely absorbed. The specific reaction
depends on the type and energy of the bombarding particle as well as the composition of the target. In either case, target nuclei are left in excited states, and this excitation energy is disposed of by the emission of particulate (protons and neutrons)
and electromagnetic (gamma rays) radiations. Gallium-67 (Ga-67) is an example
of a cyclotron-produced radionuclide. The production reaction is written as
follows:
Zn (p,2n) 67Ga
68
[16-1]
where the target material is zinc-68 (Zn-68), the bombarding particle is a proton (p) accelerated to approximately 20 MeV, two neutrons (2n) are emitted, and
Ga-67 is the product radionuclide. In some cases, the nuclear reaction produces
a radionuclide that decays to the clinically useful radionuclide (see iodine-123
and thallium-201 production below). Most cyclotron-produced radionuclides
are neutron poor and therefore decay by positron emission or electron capture.
The production methods of several cyclotron-produced radionuclides important
to nuclear medicine are shown below, (EC electron capture, T physical
half-life).
Iodine-123 production:
127
124
EC
I (p,5n)123 Xe
123 I or
T1/2 2 hr
EC or +
EC
Xe(p,2n)123 Cs
123 Xe
123 I
T 1sec
T 2hr
12
[16-2]
12
Indium-111 production:
Ag (,2n) 111In or
109
Cd (p,n) 111In or
111
Cd (p,2n) 111In
112
Cobalt-57 production:
Fe (d,n) 57Co
56
Thallium-201 production:
203
EC or b
Tl (p,3n) 201 Pb
201 Tl
T1/2 9.4 hr
Industrial cyclotron facilities that produce large activities of clinically useful

radionuclides are very expensive and require substantial cyclotron and radiochemistry support staff and facilities. Cyclotron-produced radionuclides are usually more
expensive than those produced by other technologies.
Much smaller, specialized cyclotrons, installed in commercial radiopharmacies

serving metropolitan areas or in hospitals, have been developed to produce positron-emitting radionuclides for positron emission tomography (PET) (Fig. 16-3).
These cyclotrons operate at lower energies (10 to 30 MeV) than industrial cyclotrons and commonly accelerate H ions. (An H ion is a proton with two orbital
electrons.) In such a cyclotron, the beam is extracted by passing it through a carbon
stripping foil, which removes the electrons thus creating an H ion (proton) beam.
FIGURE 16-3 Commercial self-shielded cyclotron for radionuclide production capable of producing a 60 A
beam of protons accelerated to ~11 MeV is shown with the radiation shields closed (A). Designed with a small
footprint to fit into a relatively small room (24 23 14 height) (B) Power supply and control cabinet (C)
Cyclotron assembly approximately 10,000 kg (22,000lb) (D) Retractable radiation shielding (open) approximately 14,500 kg (32,000 lb) of borated concrete and polyethylene. Neutrons and gamma radiation are an
unavoidable byproduct of the nuclear reactions which are used to produce the desired radioactive isotopes.
Boron and polyethylene are added to the radiation shield to absorb neutrons. The shielding is designed so that
radiation exposure rates are reduced to the point where technologists and other radiation workers can occupy
the room while the accelerator is in operation (less than 20 uSv/h at 24 ft from the center of the cyclotron).
(E) Cyclotron assembly open. Hydrogen gas line at the top of the cyclotron assembly provides the source of
hydrogen ions to be accelerated. (F) One of four cyclotron dees. The acceleration potential is supplied by high
frequency voltage. In this system, four dees provide eight accelerations per orbit, thus reducing acceleration
path length and beam loss. (G) Beam shaping magnets act as powerful lenses to confine ions to the midplane.
The dotted white arrow shows the beam path through one of the dees.
The radiochemicals produced (in gas or liquid) are sent through tubing in a shielded channel running
under the floor to the automated radiochemistry unit located in a shielded enclosure in a room next to the
cyclotron. A typical production run from a cyclotron in a commercial radiopharmacy serving a metropolitan
area will produce approximately 131 GBq (3.5 Ci) of F-18 during a 2 h irradiation. The radiopharmacy may
have three to four production runs a day depending on the clinical demand in the area. (Photos courtesy of
Siemens Medical Solutions, Inc.)
598
Because of the change in the polarity of the charge on each particle, the direction
of the forces on the moving particles from the magnetic field is reversed and the
beam is diverted out of the cyclotron and onto a target. These commercially available specialized medical cyclotrons have a number of advantages, including automated cyclotron operation and radiochemistry modules, allowing a technologist
with proper training to operate the unit. Radiation shielding of cyclotrons is always
an important consideration; however, the use of negative ions avoids the creation
of unwanted radioactivity in the cyclotron housing and thus reduces the amount
of radiation shielding necessary. This substantially reduces the size and weight of
the cyclotron facility allowing it to be placed within the hospital close to the PET
imaging facilities. Production methods of clinically useful positron-emitting radionuclides are listed below.
Fluorine-18 production: 18O (p,n) 18F
Nitrogen-13 production: 16O(p,a) 13N
(T1/2 110 min)

(T1/2 10 min)
Oxygen-15 production: 14N (d,n) 15O or 15N (p,n)15O (T1/2 2.0 min)
Carbon-11 production: 14N (p,a) 11C
(T1/2 20.4 min)
[16-3]
In the interests of design simplicity and cost, some medical cyclotrons accelerate
only protons. These advantages may be offset for particular productions such as 15O
when an expensive rare isotope 15N that requires proton bombardment must be used
in place of the cheap and abundant 14N isotope that requires deuteron bombardment.
The medical cyclotrons are usually located near the PET imaging system because of
the short half-lives of the radionuclides produced. Fluorine-18 (F-18) is an exception
to this generalization because of its longer half-life (110 min).
Nuclear ReactorProduced Radionuclides

Nuclear reactors are another major source of clinically used radionuclides.
Neutrons, being uncharged, have an advantage in that they can penetrate the
nucleus without being accelerated to high energies. There are two principal
methods by which radionuclides are produced in a reactor: nuclear fission and
neutron activation.
Nuclear Fission
Fission is the splitting of an atomic nucleus into two smaller nuclei. Whereas some
unstable nuclei fission spontaneously, others require the input of energy to overcome
the nuclear binding forces. This energy is often provided by the absorption of neutrons. Neutrons can induce fission only in certain very heavy nuclei. Whereas highenergy neutrons can induce fission in several such nuclei, there are only three nuclei
of reasonably long half-life that are fissionable by neutrons of all energies; these are
called fissile nuclides.
The most widely used fissile nuclide is uranium-235 (U-235). Elemental uranium exists in nature primarily as U-238 (99.3%) with a small fraction of U-235
(0.7%). U-235 has a high fission cross section (i.e., high fission probability); therefore, its concentration is usually enriched (typically to 3% to 5%) to make the fuel
used in nuclear reactors.
When a U-235 nucleus absorbs a neutron, the resulting nucleus (U-236) is in
an extremely unstable excited energy state that usually promptly fissions into two
smaller nuclei called fission fragments. The fission fragments fly apart with very high
kinetic energies, with the simultaneous emission of gamma radiation and the ejection
of two to five neutrons per fission (Equation 16-4).
134
50
Sn
235
92
U + n thermal
1
0
235
92
+3 01 n fast + g + ~200 MeV
99
42
Mo
The fission of uranium creates fission fragment nuclei having a wide range of mass
numbers. More than 200 radionuclides with mass numbers between 70 and 160 are
produced by the fission process (Fig. 16-4). These fission products are neutron rich
and therefore almost all of them decay by beta-minus () particle emission.
Nuclear Reactors and Chain Reactions

The energy released by the nuclear fission of a uranium atom is more than 200 MeV.
Under the right conditions, this reaction can be perpetuated if the fission neutrons
interact with other U-235 atoms, causing additional fissions and leading to a selfsustaining nuclear chain reaction (Fig. 16-5). The probability of fission with U-235
is greatly enhanced as neutrons slow down or thermalize. The neutrons emitted from
fission are very energetic (called fast neutrons), and are slowed (moderated) to thermal
energies (~0.025 eV) as they scatter in water in the reactor core. Good moderators
are low-Z materials that slow the neutrons without absorbing a significant fraction of
them. Water is the most commonly used moderator, although other materials, such
as graphite (used in the reactors at the Chernobyl plant in the Ukraine) and heavy
water (2H2O), are also used.
Some neutrons are absorbed by nonfissionable material in the reactor, while
others are moderated and absorbed by U-235 atoms and induce additional fissions.
The ratio of the number of fissions in one generation to the number in the previous
10
90Sr
FIGURE 16-4 Fission yield as a percentage of total

fission products from uranium 235.
137Cs
99Mo
133Xe
131I
Fission Yield (%)
101
102
103
80
100
120
140
Mass Number
160
600
U235
on
utr
Ne
FIGURE 16-5 Schematic of a nuclear chain reaction. The neutrons (shown as small blackened circles) are
not drawn to scale with respect to the uranium atoms.
generation is called the multiplication factor. When the number of fissions per generation is constant, the multiplication factor is 1 and the reactor is said to be critical. When the multiplication factor is greater than 1, the rate of the chain reaction
increases, at which time the reactor is said to be supercritical. If the multiplication
factor is less than 1 (i.e., more neutrons being absorbed than produced), the reactor
is said to be subcritical and the chain reaction will eventually cease.
This chain reaction process is analogous to a room whose floor is filled with
mousetraps, each one having a ping-pong ball placed on the trap. Without any form
of control, a self-sustaining chain reaction will be initiated when a single ping-pong
ball is tossed into the room and springs one of the traps. The nuclear chain reaction
is maintained at a desired level by limiting the number of available neutrons through
the use of neutron-absorbing control rods (containing boron, cadmium, indium, or a
mixture of these elements), which are placed in the reactor core between the fuel elements. Inserting the control rods deeper into the core absorbs more neutrons, reducing the reactivity (i.e., causing the neutron flux and power output to decrease with
time). Removing the control rods has the opposite effect. If a nuclear reactor accident
results in loss of the coolant, the fuel can overheat and melt (so-called meltdown).
However, because of the design characteristics of the reactor and its fuel, an atomic
explosion, like those from nuclear weapons, is impossible.
Figure 16-6 is a diagram of a typical radionuclide production reactor. The fuel is
processed into rods of uranium-aluminum alloy approximately 6 cm in diameter and
2.5 m long. These fuel rods are encased in zirconium or aluminum, which have favorable neutron and heat transport properties. There may be as many as 1,000 fuel rods
in the reactor, depending on the design and the neutron flux requirements. Water
Concrete
shielding
Reactor core
Fuel rods
Radiation detectors
Water-filled
steel thermal shield
Neutron beam holes

for sample irradiation
Heated water out
Steel & masonite

thermal shield
Cooled water in
Heat exchanger
Water pump
H2O
From cooling tower/pond
Metal fuel rods
~1 x 12
Enriched U235
fuel elements
~1/2 x 8 cylinders
To cooling tower/pond
Pump motor
Reactor coolant closed loop

Heat exchanger closed loop
Cooled
Heated
H2O
Heat transfer
Fast neutrons
90Br
90Rb
90Kr
1s
33s
90Sr
2.9m
90Y
27.7y
90Zr
Stable
2.7d
n
236
(
Thermal neutron
235
U
U + n)
Fission
products
Kinetic
energy
Heat

Gamma
rays
Decay
products
n
143
143
Xe
1s
143
Cs
143
Ba
2s
143
La
12s
143
Ce
14m
Pr
1.4d
143
Nd Stable
14.6d
FIGURE 16-6 NRU Radionuclide research/production reactor (adapted from diagram provided courtesy of
Atomic Energy of Canada and Chalk River Laboratories, Chalk River, Ontario). Fuel rod assemblies and the fission process are illustrated to show some of the detail and the relationships associated with fission-produced
radionuclides.
circulates between the encased fuel rods in a closed loop, whereby the heat generated from the fission process is transferred to cooler water in the heat exchanger.
The water in the reactor and in the heat exchanger is in closed loops that do not
come into direct physical contact with the fuel. The heat transferred to the cooling
water is released to the environment through cooling towers, evaporation ponds, or
heat exchangers that transfer the heat to a large body of water. The cooled water is
pumped back toward the fuel rods, where it is reheated and the process is repeated.
In commercial nuclear power electric generation stations, the heat generated
from the fission process produces high pressure steam that is directed through a
steam turbine, which powers an electrical generator. The steam is then condensed to
water by the condenser.
602
Nuclear reactor safety design principles dictate numerous barriers between the
radioactivity in the core and the environment. For example, in commercial power reactors, the fuel is encased in metal fuel rods that are surrounded by water and enclosed
in a sealed, pressurized, approximately 30-cm-thick steel reactor vessel. These components, together with other highly radioactive reactor systems, are enclosed in a
large steel-reinforced concrete shell (~1 to 2 m thick), called the containment structure.
In addition to serving as a moderator and coolant, the water in the reactor acts as a
radiation shield, reducing the radiation levels adjacent to the reactor vessel. Specialized nuclear reactors are used to produce clinically useful radionuclides from fission
products or neutron activation of stable target material.
Fission-Produced Radionuclides
The fission products most often used in nuclear medicine are molybdenum-99 (Mo99), iodine-131 (I-131), and xenon-133 (Xe-133). These products can be chemically
separated from other fission products with essentially no stable isotopes (carrier) of
the radionuclide present. Thus, the concentration or specific activity (measured in
MBq or Ci per gram) of these carrier-free fission-produced radionuclides is very
high. High-specific-activity, carrier-free nuclides are preferred in radiopharmaceutical preparations to increase the labeling efficiency of the preparations and minimize
the mass and volume of the injected material.
Neutron ActivationProduced Radionuclides

Neutrons produced by the fission of uranium in a nuclear reactor can be used to
create radionuclides by bombarding stable target material placed in the reactor. Ports
exist in the reactor core between the fuel elements where samples to be irradiated
are inserted. This process, called neutron activation, involves the capture of neutrons
by stable nuclei, which results in the production of radioactive nuclei. The most
common neutron capture reaction for thermal (slow) neutrons is the (n,) reaction,
in which the capture of the neutron by a nucleus is immediately followed by the
emission of a gamma ray. Other thermal neutron capture reactions include the (n,p)
and (n,) reactions, in which the neutron capture is followed by the emission of a
proton or an alpha particle, respectively. However, because thermal neutrons can
induce these reactions only in a few, low-atomic-mass target nuclides, most neutron
activation uses the (n,) reaction. Almost all radionuclides produced by neutron activation decay by beta-minus particle emission. Examples of radionuclides produced
by neutron activation useful to nuclear medicine are listed below.
Phosphorus-32 production: 31P(n,) 32P

Chromium-51 production: 50Cr(n,) 51Cr
(T1/2 14.3 days)

[16-5]
(T1/2 27.8 days)
A radionuclide produced by an (n,) reaction is an isotope of the target element.

As such, its chemistry is identical to that of the target material, making chemical
separation techniques useless. Furthermore, no matter how long the target material is irradiated by neutrons, only a small fraction of the target atoms will undergo
neutron capture and become activated. Therefore, the material removed from the
reactor will not be carrier free because it will always contain stable isotopes of the
radionuclide. In addition, impurities in the target material will cause the production of other radionuclides. The presence of carrier in the mixture limits the ability
to concentrate the radionuclide of interest and therefore lowers the specific activity.
For this reason, many of the clinically used radionuclides that could be produced
by neutron activation (e.g., 131I, 99Mo) are instead produced by nuclear fission to
TABLE 16-1 COMPARISON OF RADIONUCLIDE PRODUCTION METHODS

PRODUCTION METHOD
LINEAR
ACCELERATOR/
CYCLOTRON
NUCLEAR
REACTOR
(FISSION)
NUCLEAR REACTOR
(NEUTRON
ACTIVATION)
Bombarding
particle
Proton, alpha
Neutron
Neutron
Production by
decay of parent
Product
Neutron poor
Neutron
excess
Neutron excess
Neutron poor
or excess
Typical decay
pathway
Positron
emission,
electron capture
Beta-minus
Beta-minus
Several modes
Typically carrier
free
Yes
Yes
No
Yes
High specific
activity
Yes
Yes
No
Yes
Relative cost
High
Low
Low
Low (99mTc)
Radionuclides
for nuclear
medicine
applications
11
CHARACTERISTIC
RADIONUCLIDE
GENERATOR
High (82Rb)
C, N, O, F,
Co, 67Ga, 68Ge,
111
In, 123I, 201Tl
13
15
18
99
Mo,
57
I,
131
Xe
133
32
P, Cr, Sr, I,
Sm
51
89
125
153
68
90
Ga, 81mKr, 82Rb,

Y, 99mTc
maximize specific activity. An exception to the limitations of neutron activation is the

production of 125I, in which neutron activation of the target material, 124Xe, produces
a radioisotope, 125Xe, that decays to form the desired radioisotope (Equation 16-6).
In this case, the product radioisotope can be chemically or physically separated from
the target material. Various characteristics of radionuclide production are compared
in Table 16-1.
I production:
125
124
EC or b
Xe (n,g )125 Xe
125 I
T1/2 17 hr
[16-6]
Radionuclide Generators
Since the mid-1960s, technetium-99m (Tc-99m) has been the most important radionuclide used in nuclear medicine for a wide variety of radiopharmaceutical applications. However, its relatively short half-life (6 h) makes it impractical to store even
a weekly supply. This supply problem is overcome by obtaining the parent Mo-99,
which has a longer half-life (67 h) and continually produces Tc-99m. The Tc-99m
is collected periodically in sufficient quantities for clinical operations. A system for
holding the parent in such a way that the daughter can be easily separated for clinical
use is called a radionuclide generator.
Molybdenum-99/Technetium-99m Radionuclide Generator

In a molybdenum-99/technetium-99m radionuclide generator, Mo-99 (produced by
nuclear fission of U-235 to yield a high-specific-activity, carrier-free parent) is loaded,
in the form of ammonium molybdenate (NH4)(MoO4), onto a porous column
604
containing 5 to 10 g of an alumina (Al2O3) resin. The ammonium molybdenate

becomes attached to the surface of the alumina molecules (a process called adsorption). The porous nature of the alumina provides a large surface area for adsorption
of the parent.
As with all radionuclide generators, the chemical properties of the parent and daughter are different. In the Mo-99/Tc-99m or moly generator, the Tc-99m is much less
tightly bound than the Mo-99. The daughter is removed (eluted) by the flow of isotonic
(normal, 0.9%) saline (the eluant) through the column. When the saline solution is
passed through the column, the chloride ions easily exchange with the TcO4 (but not
the MoO4) ions, producing sodium pertechnetate, Na (99mTcO4). Technetium-99m
pertechnetate (99mTcO4) is produced in a sterile, pyrogen-free form with high specific
activity and a pH (~5.5) that is ideally suited for radiopharmaceutical preparations.
Commercially moly generators have a large reservoir of oxygenated saline
(the eluant) connected by tubing to one end of the column and a vacuum extraction
vial to the other. On insertion of the vacuum collection vial (contained in a shielded
elution tool), saline is drawn through the column and the eluate is collected during
elution which takes about 1 to 2 min. Figure 16-7 is a picture and cross-sectional
diagram of a moly generator together with an insert that shows details of the generator
column. Sterility is achieved by a millipore filter connected to the end of the column,
by the use of a bacteriostatic agent in the eluant, or by autoclave sterilization of the
loaded column by the manufacturer.
Moly generators are typically delivered with approximately 37 to 740 GBq (1 to
20 Ci) of Mo-99, depending on the workload of the department. The larger activity
generators are typically used by commercial radiopharmacies supplying radiopharmaceuticals to multiple nuclear medicine departments. The generators are shielded
by the manufacture with lead, tungsten or in the case of higher activity generators
depleted uranium. Additional shielding is typically placed around the generator to
reduce the exposure of staff during elution. The activity of the daughter at the time
of elution depends on the following:
1. The activity of the parent
2. The rate of formation of the daughter, which is equal to the rate of decay of the
_
parent (i.e., A o e l p t )
3. The decay rate of the daughter
4. The time since the last elution
5. The elution efficiency (typically 80% to 90%)
Transient Equilibrium
Between elutions, the daughter (Tc-99m) builds up or grows in as the parent (Mo99) continues to decay. After approximately 23 h, the Tc-99m activity reaches a maximum, at which time the production rate and the decay rate are equal and the parent
and daughter are said to be in transient equilibrium. Once transient equilibrium has
been achieved, the daughter activity decreases, with an apparent half-life equal to the
half-life of the parent. Transient equilibrium occurs when the half-life of the parent
is greater than that of the daughter by a factor of approximately 10. In the general
case of transient equilibrium, the daughter activity will exceed the parent at equilibrium. If all of the (Mo-99) decayed to Tc-99m, the Tc-99m activity would slightly
exceed (~10% higher) that of the parent at equilibrium. However, approximately
12% of Mo-99 decays directly to Tc-99 without first producing Tc-99m, Figure
16-8. Therefore, at equilibrium, the Tc-99m activity will be only approximately 97%
(1.1 0.88) that of the parent (Mo-99) activity.
Eluant
Vial
Evacuated
Vial
Alignment Adapter
Elution
Shield Handle
Needle Cap
Needle Guard
Main
Shielding
body
Generator
Auxiliary Shield
Elution Shield Cap

Top and Bottom
Shielding
(Tungsten)
Elution Tool
Eluate
99mTcO
4
Eluant
0.9% NaCI
Cross section of a generator
Plastic enclosure
NaCI
Air filter
Eluant inlet needle

Rubber stopper
Lead
shielding
Disperses
saline to
obtain
Silica gel
maximum
Band of 99Mo 99mTc yield
Glass wool
Column
Plastic
case
Alumina absorbs
free 99Mo to keep
breakthrough
to a minimum
Alumina
(aluminum oxide)
Glass filter (retains
Al oxide particulates)
Outlet needle
to collection
vial (eluate)
Rubber stopper
Detail of column
B
FIGURE 16-7 (A) Picture of a wet molybdenum 99/technetium 99m generator in the process of being
eluted (left). A spent generator which is no longer radioactive was used in order to minimize dose. For picture
clarity, the shielding normally surrounding the generator (illustrated in the accompanying diagram (right) is
not shown. However correct radiation safety principles (discussed further in chapter 21) are shown including
the use of disposable gloves, finger ring and body dosimeters and disposable plastic backed absorbent paper
on the bench top to minimize the spread of any contamination. An explosion diagram depicting the generator
components, and auxiliary radiation shielding is shown on the right. (B) A cross-sectional diagram of the generator interior and column detail. Consult the text for additional information on the elution process. (Adapted
from photo and diagrams provided courtesy of Covidien Radiopharmaceuticals St. Louis, MO.)
605
606
Mo-99 (T 66 hr)
88%
12%
Tc-99 m
Tc-99
(T 6.02 hr)
5
(T 2.12 10 yr)
Ru-99 (stable)
FIGURE 16-8 Simplified decay scheme of Mo-99. Mo-99 decays to Tc 99m approximately 88% of the
time. Thus is due to the 8 transition directly to Tc-99m (,82.2%) along with several other beta transitions
to excited states that emit gamma rays (principally the 4 10 and 7 4) to yield Tc-99m. The balance (12%)
of Mo-99 decays occurs by other beta transitions to excited states that ultimately yield Tc-99 bypassing the
metastable form of Tc (Tc-99m).
Moly generators (sometimes called cows) are usually delivered weekly and
eluted (called milking the cow) each morning, allowing maximal yield of the
daughter. The elution process is approximately 90% efficient. This fact, together
with the limitations on Tc-99m yield in the Mo-99 decay scheme, results in a maximum elution yield of approximately 85% of the Mo-99 activity at the time of elution.
Therefore, a typical elution on Monday morning from a moly generator with 55.5
GBq (1.5 Ci) of Mo-99 yields approximately 47.2 GBq (1.28 Ci) of Tc-99m in 10
mL of normal saline (a common elution volume). By Friday morning of that week,
the same generator would be capable of delivering only about 17.2 GBq (0.47 Ci).
The moly generator can be eluted more frequently than every 23 h; however, the
Tc-99m yield will be less. Approximately half of the maximal yield will be available
6 h after the last elution. Figure 16-9 shows a typical time-activity curve for a moly
generator.
Secular Equilibrium
Although the moly generator is by far the most widely used in nuclear medicine,
other generator systems produce clinically useful radionuclides. When the half-life
of the parent is much longer than that of the daughter (i.e., more than about 100
times longer), secular equilibrium occurs after approximately five to six half-lives of
the daughter. In secular equilibrium, the activity of the parent and the daughter
are the same if all of the parent atoms decay directly to the daughter. Once secular
equilibrium is achieved, the daughter will have an apparent half-life equal to that
of the parent. The strontium-82/rubidium-82 (Sr-82/Rb-82) generator, with parent
and daughter half-lives of 25.5 d and 75 s, respectively, reaches secular equilibrium
within approximately 7.5 min after elution. Figure 16-10 shows a time-activity curve
demonstrating secular equilibrium. The characteristics of radionuclide generator systems are compared in Table 16-2.
% of Mo-99 Activity @ T = 0
Elution @ Max Tc-99m Activity ~97% of

Mo-99 Activity @ max in-growth
100
90
80
70
60
Mo-99 Activity
In-Growth 99mTc
50
40
Elution
(~85% efficiemcy)
30
20
~23 hrs
~21 hrs
~21 hrs
10
0
16
24
32
40
48
56
Time (hours)
FIGURE 16-9 Time-activity curve of a molybdenum 99/technetium 99m radionuclide generator system
demonstrating the ingrowth of Tc-99m and subsequent elution. The time to maximum Tc-99m activity,
approximately 23 h, assumes there is no residual Tc-99m from a previous elution of the column. Typical elution
efficiency is approximately 85% (15% residual Tc-99m), thus time to maximum Tc-99m activity following the
first elution is approximately 21 h. Approximately 50% of the maximum Tc-99m activity is obtained in 6 h. The
maximum Tc-99m activity in the eluate is typically 80% to 90% of Mo-99 activity.
Quality Control
The users of moly generators are required to perform molybdenum and alumina
breakthrough tests. Mo-99 contamination in the Tc-99m eluate is called molybdenum breakthrough. Mo-99 is an undesirable contaminant because its long half-life
TABLE 16-2CLINICALLY USED RADIONUCLIDE GENERATOR SYSTEMS IN
NUCLEAR MEDICINE
PARENT
DECAY MODE
AND (HALF-LIFE)
DAUGHTER
TIME OF
MAXIMAL
INGROWTH
(EQUILIBRIUM)
DECAY MODE
AND (HALF-LIFE)
DECAY PRODUCT
68
Ge
EC (271 d)
68
Ga
6.5 h (S)
EC (68 min)
68
Zn (stable)
90
Sr
(28.8 y)
90
1 mo (S)
(2.67 d)
90
Zr (stable)
81
Rb
EC (4.6 h)
81m
80 s (S)
IT (13.5 s)
81
Kra
Sr
EC (25.5 d)
82
7.5 min (S)
(75 s)
82
Kr (stable)
Mo
(67 h)
99m
24 h (T)
IT (6 h)
99
Tca
82
99
Kr
Rb
Tc
Note: Decay modes: EC, electron capture; , positron emission; , beta-minus; IT, isometric transition (i.e.,
gamma ray emission). Radionuclide equilibrium; T, transient; S, secular.
a
These nuclides have half-lives greater than 105 y and for medical applications can be considered to be
essentially stable.
608
FIGURE 16-10 Time-activity curve

demonstrating secular equilibrium.
100
Parent
A1
87.5
A2
Daughter
Activity (Percent)
75
50
1T1/2
2T1/2
3T1/2
Time (hours)
and highly energetic betas increase the radiation dose to the patient without providing any clinical information. The high-energy gamma rays (~740 and 780 keV) are
very penetrating and cannot be efficiently detected by scintillation cameras. The U.S.
Pharmacopeia (USP) and the U.S. Nuclear Regulatory Commission (NRC) limit the
Mo-99 contamination to 0.15 mCi of Mo-99 per mCi of Tc-99m at the time of administration. The Mo-99 contamination is evaluated by placing the Tc-99m eluate in a
thick (~6 mm) lead container (provided by the dose calibrator manufacturer), which
is placed in the dose calibrator. The high-energy photons of Mo-99 can be detected,
whereas virtually all of the Tc-99m 140-keV photons are attenuated by the lead container. Eluates from moly generators rarely exceed permissible Mo-99 contamination limits. The quality control procedures to evaluate breakthrough of radionuclidic
contaminates in the eluates from Mo-99/Tc-99m and Sr-82/Rb-82 generators are discussed further in chapter 17 in the context of dose calibrator operations and quality
control. It is also possible (although rare) for some of the alumina from the column
to contaminate the Tc-99m eluate. Alumina interferes with the preparation of some
radiopharmaceuticals (especially sulfur colloid and Tc-99m-labeled red blood cell
preparations). The USP limits the amount of alumina to no more than 10 mg alumina
per mL of Tc-99m eluate. Commercially available paper test strips and test standards
are used to assay for alumina concentrations.
16.2 Radiopharmaceuticals
Characteristics, Applications, Quality Control, and

Regulatory Issues in Medical Imaging
The vast majority of radiopharmaceuticals in nuclear medicine today use Tc-99m as
the radionuclide. Most Tc-99m radiopharmaceuticals are easily prepared by aseptically injecting a known quantity of Tc-99m pertechnetate into a sterile vial containing the lyophilized (freeze-dried) pharmaceutical. The radiopharmaceutical complex
is, in most cases, formed instantaneously and can be used for multiple doses over a
period of several hours. Radiopharmaceuticals can be prepared in this fashion (called

kits) as needed in the nuclear medicine department, or they may be delivered to the
department by a centralized commercial radiopharmacy that serves several hospitals
in the area. Although most Tc-99m radiopharmaceuticals can be prepared rapidly
and easily at room temperature, several products (e.g., Tc-99m macroaggregated
albumin [MAA]), require multiple steps such as boiling the Tc-99m reagent complex
for several minutes. In almost all cases, however, the procedures are simple and the
labeling efficiencies are very high (typically greater than 95%).
Other radionuclides common to diagnostic nuclear medicine imaging include
123 67
I, Ga, 111In, 133Xe, and 201Tl. Positron-emitting radionuclides are used for PET. 18F,
as fluorodeoxyglucose (FDG), is used in approximately 85% of all clinical PET applications. Rubidium 82 (82Rb) is used to assess myocardial perfusion using PET/CT
imaging systems, in place of Tl-201 and Tc-99m based myocardial perfusion agents
that are imaged using scintillation cameras. A wide variety of other positron-emitting
radionuclides are currently being evaluated for their clinical utility, including carbon-11 (11C), nitrogen-13 (13N), oxygen-15 (15O), and gallium-68 (68Ga). The physical characteristics, most common modes of production, decay characteristics, and
primary imaging photons (where applicable) of the radionuclides used in nuclear
medicine are summarized in Table 16-3.
Ideal Diagnostic Radiopharmaceuticals

Although there are no truly ideal diagnostic radiopharmaceuticals, it is helpful to
think of currently used agents in light of the ideal characteristics.
Low Radiation Dose

It is important to minimize the radiation exposure to patients while preserving the
diagnostic quality of the image. Radionuclides can be selected that have few particulate emissions and a high abundance of clinically useful photons. Most modern
scintillation cameras are optimized for photon energies close to 140 keV, which is a
compromise among patient attenuation, spatial resolution, and detection efficiency.
Photons whose energies are too low are largely attenuated by the body, increasing the
patient dose without contributing to image formation. High-energy photons are more
likely to escape the body but have poor detection efficiency and easily penetrate collimator septa of scintillation cameras (see Chapter 18). A radiopharmaceutical should
have an effective half-life long enough to complete the study with an adequate concentration in the tissues of interest but short enough to minimize the patient dose.
High Target/Nontarget Activity

The ability to detect and evaluate lesions depends largely on the concentration of
the radiopharmaceutical in the organ, tissue or, lesion of interest or on a clinically
useful uptake and clearance pattern. Maximum concentration of the radiopharmaceutical in the target tissues of interest while minimizing the uptake in surrounding (nontarget) tissues and organs improves contrast and the ability to detect subtle
abnormalities in the radiopharmaceuticals distribution. Maximizing this target/
nontarget ratio is characteristic of all clinically useful radiopharmaceuticals and is
improved by observing the recommended interval between injection and imaging
for the specific agent. This interval is a compromise between the uptake of the activity in target tissue, washout of the activity in the background (nontarget) tissues and
practical considerations of clinic operations. With some radiopharmaceuticals such
as the bone scanning agent Tc-99m labeled methylene-diphosphonate (99m Tc-MDP)
Neutron
activation
Cyclotron
produced
Cyclotron
produced
Cyclotron
produced
Cyclotron
produced
Neutron
activation
Generator
product
Nuclear
fission (235U)
Generator
product
Chromium-51 (51Cr)
Cobalt-57 (57Co)
Gallium-67 (67Ga)
Indium-111 (111In)
Iodine-123 (123I)
Iodine-125 (125I)
Krypton-81m (81mKr)
Molybdenum-99 (99Mo)
Technetium-99m (99mTc)
RADIONUCLIDE
METHOD OF
PRODUCTION
IT (100)
(100)
IT (100)
EC (100)
EC (100)
EC (100)
EC (100)
EC (100)
EC (100)
MODE OF
DECAY (%)
140 (88)
740 (12)
780 (4)
190 (67)
181 (6)
740 (12)
35 (6)
27 (39) XR
28 (76) XR
31 (20) XR
159 (83)
171 (90)
245 (94)
93 (40)
184 (20)
300 (17)
393 (4)
122 (86)
136 (11)
320 (9)
PRINCIPAL
PHOTONS keV
(% ABUNDANCE)
6.02 h
67 h
13 s
60.2 d
13.2 h
2.8 d
78 h
271 d
27.8 d
PHYSICAL
HALF-LIFE
This radionuclide accounts for more than 70% of all imaging

studies.
Parent material for Mo/Tc generator. Not used directly as a

radiopharmaceutical; 740- and 780-keV photons used to identify
moly breakthrough.
This ultrashort-lived generator-produced radionuclide is a gas

and can be used to perform serial lung ventilation studies with
very little radiation exposure to patient or staff. The expense
and short T1/2 of the parent (81Rb, 4.6 h) limits its use.
Typically used as 125I albumin for in vivo blood/plasma volume

determinations (not used for imaging; samples counted in a
NaI(Tl) well counter).
Has replaced 131I for diagnostic imaging to reduce patient

radiation dose.
Typically used when the kinetics require imaging more than

24 h after injection. Both photons are used in imaging.
Typically use the 93, 184, and 300 keV photons for imaging.
Principally used as a uniform flood field source for scintillation

camera quality control.
Used for in vivo red cell mass determinations (not used for
imaging; samples counted in a NaI(Tl) well counter).
COMMENTS
TABLE 16-3 PHYSICAL CHARACTERISTICS OF CLINICALLY USED RADIONUCLIDES
610
Cyclotron
produced
Thallium (201Tl)
Cyclotron
produced
Cyclotron
produced
Cyclotron
produced
Cyclotron
produced
Generator
product
Generator
product
Carbon-11 (11C)
Fluorine-18 (18F)
Nitrogen-13 (13N)
Oxygen-15 (15O)
Gallium-68 (68Ga)
Rubidium-82 (82Rb)
Positron-emitting radionuclides
Nuclear
fission (235U)
Xenon-133 (133Xe)
(95) EC (5)
(89) EC (11)
(99.9)
(99.8)
(97) EC (3)
(99.8)
EC (100)
(100)
511 AR (190)
776 (13)
511 AR (184)
511 AR (200)
511 AR (200)
511 AR (193)
511 AR (200)
6980 (94) XR
81 (37)
75 s
68 min
122 s
10 min
110 min
20.4 min
73.1 h
5.3 d
(continued)
Rb-82 is a generator decay product of Sr-82, which is cyclotron

produced via a Rb-85 (p,4n) Sr-82 reaction. The half-life of Sr-82
is 25 d (600 h) and is in secular equilibrium with Rb-82 within
8 min after elution.
Ga-68 is a generator decay product of Ge-68, which is linear

accelerator produced via a Ga-69 (p,2n) Ge-68 reaction.
Cyclotron produced via 14N (d,n) 15O or 15N (p,n) 15O. Short half
life requires on-site cyclotron for imaging. Primarily clinical
research applications.
Cyclotron produced via 16O (p,) 13N reaction. Short half life
requires on-site cyclotron for imaging. Primarily clinical research
applications.
This radionuclide accounts for more than 7080% of all clinical

PET studies; typically formulated as FDG. Cyclotron produced via
18
O (p,n) 18F reaction.
Carbon-11 production: 14N (p,) 11C Short half life requires

on-site cyclotron for imaging. Primarily clinical research
applications.
The majority of clinically useful photons are low-energy x-rays

(6980 keV) from mercury 201 (201Hg), the daughter of 201Tl.
Although these photons are in high abundance (94%), their
low energy results in significant patient attenuation, which
is particularly difficult with female patients, in whom breast
artifacts in myocardial imaging often makes interpretation
more difficult.
Xe is a heavier-than-air gas. Low abundance and low energy

of photon reduces image resolution.
133
Neutron
activation
Neutron
activation
Neutron
activation
Neutron
activation
Generator
product
daughter of
Sr-90
Iodine-125 (125I)
Phosphorus-32 (32P)
Samarium-153 (153Sm)
Strontium-89 (89Sr)
Yttrium-90 (90Y)
(100)
(100)
(100)
(100)
EC (100)
(100)
MODE OF
DECAY (%)
None
Bremsstrahlung
x-ray imaging
is possible to
confirm delivery to
treatment region
None
103 (29)
None
35 (6)
27 (39) XR
28 (76) XR
31 (20) XR
284 (6)
364 (81)
637 (7)
PRINCIPAL
PHOTONS keV
(% ABUNDANCE)
2.67 d
50.53 d
1.93 d
14.26 d
60.2d
8.0 d
PHYSICAL
HALF-LIFE
Bound to murine monoclonal antibodies, used to treat certain

non-Hodgkins lymphomas. Also used bound to microspheres
(glass or resin) for intrahepatic arterial delivery of the Y-90
microspheres for the treatment of unresectable metastatic liver
tumors .Y-90 is a generator decay product of Sr-90, which is a
fission product.
As strontium chloride for pain relief from metastatic bone

lesions.
As 153Sm lexidronam ethylene diamine tetra methylene

phosphonic acid (EDTMPA) used for pain relief from metastatic
bone lesions. Advantage compared to 89Sr is that the 153Sm
distribution can be imaged.
Prepared as either sodium phosphate for treatment of

myeloproliferative disorders such as polycythemia vera and
thrombocytosis or colloidal chromic phosphate for intracavitary
therapy of malignant ascites, malignant pleural effusions,
malignant pericardial effusions, and malignant brain cysts.
Recent use as 125I Iotrex liquid brachytherapy source in Proxima

GliaSite radiation therapy system for treatment of recurrent
gliomas and metastatic brain tumors.
Used for treatment of hyperthyroidism and thyroid cancer:

364-keV photon used for imaging. Resolution and detection
efficiency are poor due to high energy of photons. High patient
dose, mostly from particles.
COMMENTS
Note: , Beta-minus decay; , beta-plus (positron) decay; AR, annihilation radiation; EC, electron capture; IT, isomeric transition (i.e., gamma ray emission), XR, x-ray.
Nuclear
fission (235U)
Iodine-131 (131I)
Therapeutic radionuclides
RADIONUCLIDE
METHOD OF
PRODUCTION
TABLE 16-3 PHYSICAL CHARACTERISTICS OF CLINICALLY USED RADIONUCLIDES (continued)
612
instructions to the patient to be well hydrated and void prior to imaging improves
image quality, lesion detectability and reduces patient dose. Abnormalities can be
identified as localized areas of increased radiopharmaceutical concentration, called
hot spots (e.g., a stress fracture in a bone scan), or as cold spots in which the
radiopharmaceuticals normal localization in a tissue is altered by a disease process
(e.g., perfusion defect in a lung scan with 99mTc-MAA). Disassociation of the radionuclide from the radiopharmaceutical alters the desired biodistribution, thus degrading
image quality. Good quality control over radiopharmaceutical preparation helps to
ensure that the radionuclide is bound to the pharmaceutical.
Safety, Convenience, and Cost-Effectiveness

Low chemical toxicity is enhanced by the use of high-specific-activity, carrier-free
radionuclides that also facilitate radiopharmaceutical preparation and minimize the
required amount of the isotope. For example, 3.7 GBq (100 mCi) of I-131 contains only 0.833 mg of iodine. Radionuclides should also have a chemical form, pH,
concentration, and other characteristics that facilitate rapid complexing with the
pharmaceutical under normal laboratory conditions. The compounded radiopharmaceutical should be stable, with a shelf life compatible with clinical use, and should
be readily available from several manufacturers to minimize cost.
Therapeutic Radiopharmaceuticals
Radiopharmaceuticals are also used for the treatment of a number of diseases. The
goal of radiopharmaceutical therapy is to deliver a sufficiently large dose to the target
organ, tissue, or cell type while limiting the dose to nontarget tissue to minimize deterministic effects such as bone marrow suppression and to minimize the risk of cancer.
All currently approved therapeutic radiopharmaceuticals emit beta particles, but
research continues with radiopharmaceuticals containing short-lived alpha particle
emitting radionuclides such as astatine-211(At-211) and Auger electronemitting
radionuclides such as I-125 and In-111.
Commonly used therapeutic radiopharmaceuticals include I-131 NaI for hyperthyroidism and thyroid cancer, Sr-89 chloride and Sm-153 lexidronam for relief of
pain from cancer metastatic to bone, I-131 and Y-90 labeled monoclonal antibodies
for treatment of certain non-Hodgkins lymphomas, and P-32 as sodium phosphate
for bone marrow disorders such as polycythemia vera. Radionuclides used for radiopharmaceutical therapy are listed in Table 16-3.
Radiopharmaceutical Mechanisms of Localization

Radiopharmaceutical concentration in tissue is driven by one or more of the following mechanisms: (1) compartmental localization and leakage, (2) cell sequestration,
(3) phagocytosis, (4) passive diffusion, (5) active transport, (6) capillary blockade,
(7) perfusion, (8) chemotaxis, (9) antibody-antigen complexation, (10) receptor
binding, and (11) physiochemical adsorption.
Compartmental Localization and Leakage

Compartmental localization refers to the introduction of the radiopharmaceutical
into a well-defined anatomic compartment. Examples include Xe-133 gas inhalation
into the lung, intraperitoneal instillation of P-32 chromic phosphate, and Tc-99mlabeled RBCs injected into the circulatory system. Compartmental leakage is used to
identify an abnormal opening in an otherwise closed compartment, as when labeled
RBCs are used to detect gastrointestinal bleeding.
614
Cell Sequestration
To evaluate splenic morphology and function, RBCs are withdrawn from the patient,
labeled with Tc-99m, and slightly damaged by in vitro heating in a boiling water bath
for approximately 30 min. After they have been reinjected, the spleens ability to recognize and remove (i.e., sequester) the damaged RBCs is evaluated.
Phagocytosis
The cells of the reticuloendothelial system are distributed in the liver (~85%), spleen
(~10%), and bone marrow (5%). These cells recognize small foreign substances in
the blood and remove them by phagocytosis. In a liver scan, for example, Tc-99m-labeled sulfur colloid particles (100 nm) are recognized, being substantially smaller
than circulating cellular elements, and are rapidly removed from circulation.
Passive Diffusion
Passive diffusion is simply the free movement of a substance from a region of high
concentration to one of lower concentration. Anatomic and physiologic mechanisms
exist in the brain tissue and surrounding vasculature that allow essential nutrients,
metabolites, and lipid-soluble compounds to pass freely between the plasma and
brain tissue while many water-soluble substances (including most radiopharmaceuticals) are prevented from entering healthy brain tissue. This system, called the
blood-brain barrier, protects and regulates access to the brain. Disruptions of the
blood-brain barrier can be produced by trauma, neoplasms, and inflammation.
The disruption permits radiopharmaceuticals such as Tc-99m diethylenetriaminepentaacetic acid (DTPA), which is normally excluded by the blood-brain barrier, to
follow the concentration gradient and enter the affected brain tissue.
Active Transport
Active transport involves cellular metabolic processes that expend energy to concentrate the radiopharmaceutical into a tissue against a concentration gradient and
above plasma levels. The classic example in nuclear medicine is the trapping and
organification of radioactive iodide. Trapping of iodide in the thyroid gland occurs
by transport against a concentration gradient into follicular cells, where it is oxidized
to a highly reactive iodine by a peroxidase enzyme system. Organification follows,
resulting in the production of radiolabeled triiodothyronine (T3) and thyroxine (T4).
Another example is the localization of thallium (a potassium analog) in muscle tissue. The concentration of Tl-201 is mediated by the energy-dependent Na/K ionic
pump. Nonuniform distribution of Tl-201 in the myocardium indicates a myocardial
perfusion deficit. F-18 FDG is a glucose analog that concentrates in cells that rely
upon glucose as an energy source, or in cells whose dependence on glucose increases
under pathophysiological conditions. FDG is actively transported into the cell where
it is phosphorylated and trapped for several hours as FDG-6-phosphate. The retention and clearance of FDG reflects glucose metabolism in a given tissue. FDG is used
to assist in the evaluation of malignancy in patients with known or suspected diagnoses of cancer. In addition, FDG is used to assess regional cardiac glucose metabolism
for the evaluation of hibernating myocardium (i.e., the reversible loss of systolic
function) in patients with coronary artery disease.
Capillary Blockade
When particles slightly larger than RBCs are injected intravenously, they become
trapped in the capillary beds. A common example in nuclear medicine is in the assessment of pulmonary perfusion by the injection of Tc-99m-MAA, which is trapped
in the pulmonary capillary bed. Imaging the distribution of Tc-99m-MAA provides

a representative assessment of pulmonary perfusion. The microemboli created
by this radiopharmaceutical do not pose a significant clinical risk because only a very
small percentage of the pulmonary capillaries are blocked and the MAA is eventually
removed by biodegradation.
Perfusion
Relative perfusion of a tissue or organ system is an important diagnostic element in
many nuclear medical procedures. For example, the perfusion phase of a three-phase
bone scan helps to distinguish between an acute process (e.g., osteomyelitis) and remote
fracture. Perfusion is also an important diagnostic element in examinations such as
renograms, cerebral and hepatic blood flow studies and myocardial perfusion studies.
Chemotaxis
Chemotaxis describes the movement of a cell such as a leukocyte in response to a
chemical stimulus. 111In- and 99mTc-labeled leukocytes respond to products formed
in immunologic reactions by migrating and accumulating at the site of the reaction
as part of an overall inflammatory response.
Antibody-Antigen Complexation
An antigen is a biomolecule (typically a protein) that is capable of inducing the production of, and binding to, an antibody in the body. The antibody has a strong and
specific affinity for the antigen. An in vitro test called radioimmunoassay (RIA) makes
use of the competition between a radiolabeled antigen and the same antigen in the
patients serum for antibody binding sites. RIA, developed by Berson and Yalow
in the late 1950s (Berson and Yalow 1960), led to a Nobel Prize in Medicine for
Yalow in 1977, five years after the untimely death of Berson. RIA techniques have
been employed to measure minute quantities of various enzymes, antigens, drugs,
and hormones; however, many of these tests have been replaced by immunoassays
using nonradioactive labels. At equilibrium, the more unlabeled serum antigen that
is present, the less radiolabeled antigen (free antigen) will become bound to the antibody (bound antigen). The serum level is measured by comparing the ratio between
bound and free antigen in the sample to a known standard for that particular assay.
Antigen-antibody complexation is also used in diagnostic imaging with such
agents as In-111-labeled monoclonal antibodies for the detection of colorectal carcinoma. This class of immunospecific radiopharmaceuticals promises to provide an
exciting new approach to diagnostic imaging. In addition, a variety of radiolabeled
(typically with I-131 or Y-90) monoclonal antibodies directed toward tumors are
being used in an attempt to deliver tumoricidal radiation doses. This procedure,
called radioimmunotherapy, has proven effective in the treatment of some nonHodgkins lymphomas and is under clinical investigation for other malignancies.
Receptor Binding
This class of radiopharmaceuticals is characterized by their high affinity to bind to
specific receptor sites. For example, the uptake of In-111-octreotide, used for the
localization of neuroendocrine and other tumors, is based on the binding of a somatostatin analog to receptor sites in tumors.
Physiochemical Adsorption
The localization of methylenediphosphonate (MDP) occurs primarily by adsorption in the mineral phase of the bone. MDP concentrations are significantly higher
616
in amorphous calcium than in mature hydroxyapatite crystalline structures, which

helps to explain its concentration in areas of increased osteogenic activity.
A summary of the characteristics and clinical utility of commonly used radiopharmaceuticals is provided in Appendix F-1.
Radiopharmaceutical Quality Control

Aside from the radionuclidic purity quality control performed on the 99mTc-pertechnetate generator eluate, the most common radiopharmaceutical quality control procedure is the test for radiochemical purity. The radiochemical purity assay identifies the
fraction of the total radioactivity that is in the desired chemical form. Radiochemical
impurity can occur as the result of temperature changes, presence of unwanted oxidizing or reducing agents, pH changes, or radiation damage to the radiopharmaceutical (called autoradiolysis). The presence of radiochemical impurities compromises
the diagnostic utility of the radiopharmaceutical by reducing uptake in the organ of
interest and increasing background activity, thereby degrading image quality. In addition to lowering the diagnostic quality of the examination, radiochemical impurities
unnecessarily increase patient dose.
The most common method to determine the amount of radiochemical impurity in a
radiopharmaceutical preparation is thin-layer chromatography. This test is performed by
placing a small aliquot (1 drop) of the radiopharmaceutical preparation approximately
1 cm from one end of a small rectangular paper (e.g., Whatman filter paper) or a silicacoated plastic strip. This strip is called the stationary phase. The end of the strip with
the spot of radiopharmaceutical is then lowered into a glass vial containing an appropriate solvent (e.g., saline, acetone, or 85% methanol), such that the solvent front begins
just below the spot. The depth of the solvent in the vial must be low enough, so that the
spot of radiopharmaceutical on the strip is above the solvent. The solvent will slowly
move up the strip and the various radiochemicals will partition themselves at specific
locations identified by their reference values (Rf), which ranges from 0 to 1 along the
strip, according to their relative solubilities. The reference value number is the fraction
of the total distance on the strip, (from the origin, where the spot of the radiopharmaceutical is placed to a predetermined line near the top of the strip where the solvent
front ends), traveled by a particular radiochemical. Once the solvent front has reached
the top, the strip is removed and dried. The strip is cut into sections, and the percentage of the total radioactivity on each section of the strip is assayed and recorded. The
movements of radiopharmaceuticals and their contaminants have been characterized
for several solvents. Comparison of the results with these reference values allows the
identities and percentages of the radiochemical impurities to be determined.
The two principal radiochemical impurities in technetium-labeled radiopharmaceuticals are free (i.e., unbound) Tc-99m-pertechnetate and hydrolyzed Tc-99m. The
Tc-99m radiopharmaceutical complex and its associated impurities will, depending
upon the solvent, either remain at the origin (Rf 0) or move with the solvent front
to a location near the end of the strip. For example, with Whatman 31 ET chromatography paper as the stationary phase in an acetone solvent, Tc-99m-MAA remains
at the origin and any free pertechnetate or hydrolyzed Tc-99m migrates close to the
solvent front (Rf 0.9). On the other hand, I-131 (as bound NaI) moves with the
solvent front (Rf 1) in an 85% methanol solvent, while the impurity (unbound
iodide) moves approximately 20% of the distance from the origin (Rf 0.2). These
assays are easy to perform and should be used as part of a routine radiopharmacy
quality control program and whenever there is a question about the radiochemical
integrity of a radiopharmaceutical preparation.
16.3 Internal Dosimetry

The radiation doses to patients from diagnostic imaging procedures are an important
issue and, in the absence of a medical physicist at an institution (e.g., private practice radiology), radiologists and nuclear medicine physicians are often consulted as
the local experts. Radiation dosimetry is primarily of interest because radiation dose
quantities serve as indices of the risk from diagnostic imaging procedures using ionizing radiation. Dosimetry also plays an important role in radiopharmaceutical therapy
where estimates of activity necessary to produce tumoricidal doses must be weighed
against potential radiotoxicity to healthy tissue. In nuclear medicine procedures, the
chemical form of the radiopharmaceutical, its route of administration (e.g., intravenous injection, ingestion, inhalation), the administered activity, the radionuclide, and
patient-specific disease states and pharmacokinetics determine the patient dose.
Radiopharmaceutical Dosimetry: Methodology

Several formalized methodologies have been proposed for calculation of internal
dose. The Medical Internal Radiation Dosimetry (MIRD) and Radiation Dose Assessment Resource (RADAR) task groups of the Society of Nuclear Medicine have developed formalisms for calculating the radiation dose to selected organs and the whole
body from internally administered radionuclides. These formalisms take into account
variables related to the physics of energy deposition by ionizing radiation as well
as those associated with the biologic system for which the dose is being calculated.
Although some of the variables are known with a high degree of accuracy, others
are based on estimates or simplifying assumptions that, taken together, provide an
approximation of the dose to the average (reference) adult, adolescent, child, and the
fetus. Prior to the widespread availability of microcomputers, basic standardized computational anthropomorphic phantoms were developed by utilizing a combination of
simple geometric shapes (e.g., cones, spheres, cylinders) to represent the human body
(Fig.16-11A). Over the years, greater computer processing power and more complex
calculational techniques allowed for much more sophisticated and anatomically accurate phantoms to be developed. Today, these newer phantoms are available for adult
males and females, children of different ages, and the pregnant female (Fig. 16-11B).
The internal dosimetry formalism has two main elements: (1) estimation of the
number of nuclear transformations occurring in various source organs containing
radioactive material and (2) estimation of the energy absorbed in selected target
organs for which the dose is being calculated from the radiation emitted in source organs
(Fig. 16-12).
MIRD Formalism
The MIRD formalism designates a source organ in which activity is located as rS, and a target organ for which the dose is being calculated as rT. Following the administration of the
radiopharmaceutical, the mean dose to a particular target organ or tissue over a defined
dose integration period (TD) is calculated from the following equation (Bolch 2009)
TD
D ( r T ,T D ) =
A(r , t) M(r , t) E Y f(r

S
rS
i i
rS , E i , t )
[16-7]

where A(rS,t) is the activity for each source organ (rS) at time (t); M(rT,t) is the target
organ mass at time t; Ei is the mean energy of a given radiation emission i; and Yi
618
FIGURE 16-11 (A) First-generation computational anthropomorphic phantoms (referred to as the FisherSnyder phantoms) were developed utilizing simplified geometric shapes. Monte Carlo computer programs
were used simulate the creation and transport of radiation from the decay of radionuclides inside the body.
(B) More sophisticated and anatomically correct phantoms are currently available for adult males and females,
children of different ages and the pregnant female. (Adapted from Stabin MG, Emmons-Keenan MA, Segars
WP, Fernald MJ. The Vanderbilt University Reference Adult and Pediatric Phantom Series, and Xu XG, Shi C,
Stabin MG, Taranenko V. Pregnant female/fetus computational phantoms and the latest RPI-P series representing 3, 6, and 9 months gestational periods. In: Handbook of Anatomical Models for Radiation Dosimetry. Boca
Raton, FL: CRC Press, 2009.)
is its yield (i.e., number emitted per nuclear transformation) and f(rTrS,Ei,t), the
absorbed fraction, is the fraction of the energy emitted in the source organ rS that is
absorbed in the target rT for a given radiation emission i at time t. All of the terms after
the activity term have also been consolidated into S factors, which gives the absorbed
dose in the target, rT, per disintegration in each source organ rS (rad/Ci-hr). The
MIRD committee has not published S factors in SI units (Gy/Bq-s).
If the dose to a patient is being calculated from the administration of a diagnostic
radiopharmaceuticals during which there would be no significant change in the mass
of the source or target organs over time (a situation more relevant to therapy where
the mass of a tumor may change over time), the calculation simplifies to this timeindependent form:

D ( r T ,T D ) =
(r ,T
S
rS
D ) S (rT
rS )
[16-8]
where term (rS,TD) represents the time-integrated activity (i.e., total number of disintegrations) in the source organ or tissue during the specified dose integration period
TD. The factors that influence the time-integrated activity in a source organ are discussed below.
RADAR Formalism
The last publication of complete tables of S-values for an anthropomorphic
phantom was in 1975. Since that time, the RADAR group published a method for
Target organs
Source organ
FIGURE 16-12 Illustration of source and target organ concept for calculation of the dose to the lungs
and thyroid gland (target organs) from a radiopharmaceutical (e.g., technetium-99m sulfur colloid) primarily
located in the liver (source organ). Note that the relative geometry and mass of the source and target organs
together with the physical decay characteristic of the radionuclide and its associated radiopharmaceutical
kinetics all play a part in determining the dose to any particular organ.
calculation of internal dose for which the numerous variables and complexities
inherent in such a calculation are represented by simplified terms and symbols
(Stabin,2003).

D T = Ns DF
[16-9]
Here DT is the dose in target organ T; Ns is the number of disintegrations that

occur in any source region (i.e., the time integrated activity); and DF is the dose
factor, representing the absorbed dose in the target (T) per disintegration in each
source organ, incorporating the same variables as in the MIRD S factors. This method
has been implemented in software (OLINDA/EXM) designed to run on a desktop
computer (Stabin, 2005). The software can perform dose calculation for over 800
radionuclides in six reference adult and child models (Cristy, 1987), and three models of the pregnant female (representing the end of each trimester) using absorbed
620
fractions calculated by Oak Ridge National Laboratory (Stabin, 1995). Version 2 of

this software will perform calculations using these historical phantoms as well as
the new, more mathematically sophisticated and anatomically correct image-based
models that have been developed over the last decade for adults, children, and
pregnant females.
Time-Integrated Activity
The MIRD committee used the term cumulated activity for many years to represent
the time-integral of activity in Equation 16-7. The integral of any time-activity curve
is the total number of nuclear transformations from the radionuclide located in a
particular source organ. The RADAR method uses the term Ns, to represent the
number of disintegrations occurring in a source organ. The units of Bq-s (Becquerel
seconds) express the total number of disintegrations (i.e., disintegrations per unit
time time disintegrations).
Calculating the number of disintegrations in the various organs requires fitting
time-activity data to mathematical functions (typically one or more exponentials)
or creation and solution of a theoretical compartmental model describing the
kinetics of the radiopharmaceutical in the body. A compartment may be a physical space (e.g., blood in the vasculature) or a physiological compartment (e.g.,
iodine as thyroxine). Such a model is typically created from a knowledge of anatomy, physiology and the rates of transfer among compartments based on studies
of the in-vivo behavior of the radiopharmaceutical in animals and humans. Figure
16-13 shows a compartmental model for radioactive ionic or elemental iodine.
However, for some radiopharmaceuticals, a simplified single compartment model
may provide a reasonable approximation. In this case, activity is accumulated by
an organ and gradually eliminated. The total number of disintegrations in a source
organ depends on (1) the administered activity, (2) the fraction of the administered
activity taken up by the source organ, and (3) the rate of elimination from the
source organ.
Figure 16-14 shows a simple kinetic model representing the accumulation
and elimination of radioactivity in a source organ (liver). The activity localized in
the source organ (Af) is a fraction (f) of the injected activity (A0), (i.e., Af A0
f). If we assume there is exponential biological excretion, two processes will act
to reduce the total activity in the source organ: (1) physical decay of the radionuclide, as represented by its physical half-life Tp, and (2) biologic elimination, as
represented by the biologic half-life Tb. Taken together, the reduction of activity
in a source organ with time can be expressed as an effective half-life (Te) that is
calculated as follows:

Te =
Tp Tb

Tp + Tb
[16-10]
In a manner analogous to radioactive decay, the activity remaining in a source organ

after a time (t) is

A(rS,t) = fA0 eet
where e is the effective decay constant, equal to 0.693/Te.
[16-11]
T3 (triiodothyronine)
00
13
0.
0.
k 32
56
/d
2/
k21 = 0.1144/d
(iodide ion) I
Thyroid
42
1/
0.
0.
00
7/
k 14
k01 = 2/d
k12 = 0.005/d
Urine
4
T4 (thyroxine)
k04 = 0.018/d
Feces
FIGURE 16-13 A simplified compartmental model of iodine kinetics in humans. Compartment 1 is extra
thyroidal iodide, compartment 2 is thyroidal iodine, compartment 3 is extrathyroidal T3 (triiodothyronine), and
compartment 4 is extrathyroidal T4 (thyroxine). The compartmental transfer coefficient or rate constant kxy
describes the relative rate at which iodine is transferred from compartment y to compartment x. For example
k21 is the rate constant for the transfer of extrathyroidal iodide (compartment 1) to the thyroid gland (compartment 2). Iodine also leaves the thyroid gland in the form of T3 and T4 (k32, k42) as well as some release of free
iodine back into the extrathyroidal space (k12). The relative values of the rate constants for the elimination of
iodine into the urine and feces (k01, k04) indicate that the vast majority of the iodine is eliminated in the urine.
These rate constants can be greatly altered by pathology (e.g., hyperthyroidism) or other conditions (e.g., low
iodine diet). Reference: Mones Berman, Kinetic Models for Absorbed Dose Calculations, MIRD Pamphlet No.
12, Society of Nuclear Medicine, 1977.
Just as one cannot predict the range of a single photon before it interacts, it is not
possible to know when any particular radioactive atom will decay. However, one can
describe the mean life () or average time to decay for atoms of a specific radionuclide in
a manner analogous to the mean free path (MFP) for photons discussed in Chapter 3.
Just as the average distance traveled by the photons before interacting can be
described in terms of its linear attenuation coefficient (m) or half-value layer (HVL),
(1/ or 1.44 HVL), the mean life of a nuclide is equal to 1/ or 1.44 Tp. For the
purpose of radiation dosimetry where the removal of radioactive atoms from a
source organ is more completely described by the effective half-life (Te), mean life in
the organ becomes 1/e or 1.44 Te. The total number of nuclear transformations is
equal to the area under the time-activity curve. The area under this curve (Fig.16-15)
can be shown to be equal to the product of the initial amount of activity in the source
FIGURE 16-14 Simplified kinetics model of the accumulation and elimination of radioactivity in a source organ (e.g.,
liver). A fraction (f) of the injected activity (A0) is localized in the
source organ in which the initial activity (Af) is reduced by physical decay and biologic excretion of the radiopharmaceutical.
Ao
Source organ
622
Af
organ (A0 f) and the mean life (1.44 Te). Therefore the total number of nuclear
transformations (N) can be expressed as

N 1.44 TeA0 f
[16-12]
As discussed in the RADAR dosimetry model, the dose factor (DF) is the absorbed
dose in the target (T) per disintegration in a specified source organ. The DF is
expressed in units of mGy/MBq-s, and like the S factor, it is determined by the mass
of the target organ; the quantity, type, and mean energy of the ionizing radiation
emitted per disintegration (i.e., decay scheme); and finally the fraction of the emitted
radiation energy that reaches and is absorbed by the target organ.
Each dose factor is specific to a particular source organtarget organ pair and a
specific radionuclide. Dose factors are provided in tabular form or incorporated into
computer programs for many common diagnostic and therapeutic radionuclides.
Table 16-4 lists some dose factors from the OLINDA/EXM program for Tc-99m. This
computer program can be used for calculating organ and effective doses from a wide
variety of radiopharmaceuticals used in nuclear medicine.
(A0x f) 100
90
80
70
Percent of 60
Initial Activity
Remaining 50
40
30
20
10
0
0
10
1.44 Te (8.67 hr)
15
20
Hours
25
30
35
FIGURE 16-15 Time activity curve for a hypothetical Tc-99m labeled radiopharmaceutical that has a very
long biological half-life (Tb) compared to the 6.02 h physical half-life (Tp) of Tc-99m, thus Te = Tp. The cumulated activity is equal to the area under the curve which is numerically equal to (A0.f) (1.44 Te) representing the
total number of nuclear transformations (disintegrations) in the specified source organ.
Example Dose Calculations

The simplest example of an internal dose calculation would be a situation in
which the radiopharmaceutical instantly localizes in a single organ and remains
there with an infinite Tb. Although most radiopharmaceuticals do not meet both
of these assumptions, this hypothetical example is considered for the purpose of
demonstrating the calculational technique in its simplest possible application.
Example: A patient is injected with 100 MBq of Tc-99m-sulfur colloid. Estimate the absorbed
dose to the (1) liver, (2) testes, and (3) red bone marrow.
Assumptions:
1. All of the injected activity is uniformly distributed in the liver.
2. The uptake of Tc-99m-sulfur colloid in the liver from the blood is instantaneous.
3. There is no biologic removal of Tc-99m-sulfur colloid from the liver.
In this case, the testes and red bone marrow are target organs, whereas the liver is both a source
and a target organ. The average dose to any target organ can be estimated by applying the RADAR
formalism in Equation 16-9:
DT = Ns DF
STEP 1. Calculate the number of disintegrations (Ns) in the source organ (i.e., liver) by inserting
the appropriate values into Equation 16-12 (Note that because Tb is infinite, Te Tp):
3600 s
Ns = 1.44 6 hr 1 100 MBq = 980 MBq - hr
hr
6
12
Ns = 3.1 10 MBq-s or 3.1 10 disintegrations
STEP 2. Find the appropriate dose factors for each target/source organ pair and the radionuclide
of interest (i.e., Tc-99m) from Table 16-4. The appropriate dose factors found at the
intersection of the source organs column (i.e., liver) and the individual target organs
row (i.e., liver, testes, or red bone marrow).
Dose Factor
(mGy/MBq-s)
Target (rT)
Source (rS)
Liver
Liver
3.16 106
Red bone marrow
Liver
8.32 108
Testes
Liver
1.57 109
STEP 3. Organize the assembled information in a table of organ doses:
Target organ (rT)
NS
(MBq-s)
Dose Factor
(mGy/mBq-s)
DT
(mGy)
Liver
3.1 106
3.16 106
9.8
Red bone marrow
3.1 10
8.32 10
0.26
Testes
3.1 106
1.57 109
0.0049
Compared to the dose to the liver the relatively low dose to the testes and red bone marrow is
largely due to the fact that only a small fraction of the isotropically emitted penetrating radiation
(i.e., gamma rays and x-rays) is directed toward, and absorbed by, these tissues.
4.23E-06
3.17E-09
4.18E-10
Brain
2.52E-10
4.76E-11
1.98E-08
7.46E-08
2.85E-07
LLI Wall
Small
Intestine
Stomach
Wall
2.53E-07
2.67E-07
3.41E-08
Red
Marrow
Osteogenic
Cells
Skin
1.52E-11
3.14E-08
1.09E-06
Ovaries
7.63E-09
2.21E-08
2.33E-07
1.12E-07
Lungs
Muscle
Pancreas
8.16E-10
7.24E-07
4.35E-07
Kidneys
Liver
3.97E-08
2.99E-07
1.01E-07
4.15E-10
1.58E-10
2.54E-09
9.41E-08
2.85E-07
ULI Wall
Heart Wall
3.91E-11
1.32E-11
1.54E-10
5.05E-08
3.57E-07
Breasts
Gallbladder
Wall
4.18E-10
1.80E-04
Adrenals
Brain
Adrenals
TARGET
ORGANS
3.62E-08
1.82E-07
2.01E-07
5.21E-08
1.26E-06
1.23E-07
4.50E-09
1.80E-08
7.10E-08
5.42E-09
3.10E-07
1.24E-07
7.16E-07
1.23E-05
6.49E-08
2.28E-09
1.57E-11
2.25E-08
LLI
Contents
3.01E-08
1.49E-07
1.79E-07
1.42E-07
9.23E-07
1.12E-07
1.35E-08
1.16E-07
2.13E-07
2.06E-08
1.36E-06
2.13E-07
4.22E-06
5.92E-07
4.38E-07
7.35E-09
3.91E-11
7.46E-08
SI Content
3.41E-08
1.03E-07
7.50E-08
1.23E-06
5.85E-08
9.96E-08
1.10E-07
1.47E-07
2.73E-07
2.33E-07
2.65E-07
8.53E-06
2.08E-07
9.10E-08
3.05E-07
5.73E-08
4.27E-10
2.73E-07
Stomach
Contents
3.09E-08
1.27E-07
1.43E-07
1.62E-07
7.71E-07
1.07E-07
1.77E-08
1.87E-07
2.12E-07
2.97E-08
8.37E-06
2.86E-07
1.25E-06
2.14E-07
7.53E-07
8.00E-09
4.68E-11
9.58E-08
ULI
Contents
SOURCE ORGANS
3.41E-08
1.60E-07
1.11E-07
2.65E-07
4.55E-09
8.83E-08
4.59E-07
2.13E-07
6.45E-08
5.48E-06
2.12E-08
1.66E-07
1.57E-08
4.06E-09
1.03E-07
2.41E-07
3.14E-09
2.53E-07
Heart
Contents
3.70E-08
1.60E-07
1.11E-07
3.57E-07
6.15E-09
9.20E-08
4.40E-07
2.33E-07
8.22E-08
1.19E-05
2.65E-08
2.65E-07
2.06E-08
4.90E-09
1.22E-07
2.61E-07
2.54E-09
2.85E-07
Heart
Wall
3.79E-08
1.62E-07
1.71E-07
4.97E-07
7.02E-08
9.79E-08
6.66E-08
2.93E-07
1.32E-05
8.22E-08
2.12E-07
2.53E-07
2.13E-07
5.50E-08
4.09E-07
1.99E-08
1.58E-10
7.24E-07
Kidneys
TABLE 16-4 Tc-99m DOSE FACTORS (mGy/MBq-s) FOR SOME SOURCE/TARGET ORGAN COMBINATIONS*
Liver
3.62E-08
1.24E-07
8.32E-08
3.86E-07
3.81E-08
7.52E-08
2.09E-07
3.16E-06
2.93E-07
2.33E-07
1.88E-07
1.48E-07
1.16E-07
1.44E-08
8.70E-07
6.82E-08
8.16E-10
4.35E-07
Lungs
4.02E-08
1.66E-07
1.11E-07
1.74E-07
5.39E-09
9.34E-08
3.57E-06
1.97E-07
6.66E-08
4.40E-07
1.81E-08
1.19E-07
1.35E-08
3.29E-09
7.46E-08
2.33E-07
7.63E-09
2.33E-07
624
7.55E-09
1.89E-08
1.72E-07
Urinary
Bladder
Wall
Uterus
Total Body
5.19E-10
1.25E-07
1.31E-11
6.02E-12
1.35E-07
6.88E-09
1.46E-12
6.53E-08
1.49E-07
5.17E-07
4.98E-07
2.48E-10
2.04E-09
1.40E-07
1.01E-07
1.59E-07
8.37E-07
2.12E-07
4.87E-10
4.66E-09
2.61E-08
7.83E-07
1.17E-07
5.05E-08
1.73E-08
2.62E-09
3.65E-08
2.90E-09
1.05E-07
1.41E-07
3.97E-07
1.61E-07
7.69E-10
5.43E-09
1.92E-08
1.24E-07
1.17E-07
4.87E-09
2.22E-09
5.17E-08
8.87E-07
5.16E-10
1.67E-07
1.65E-07
5.47E-09
2.17E-09
4.33E-08
7.35E-07
6.16E-10
6.63E-07
1.58E-07
6.42E-08
1.87E-08
2.95E-09
1.73E-08
3.10E-09
7.22E-08
1.59E-07
3.29E-08
1.16E-08
8.64E-09
5.93E-08
1.57E-09
1.64E-07
1.44E-07
4.10E-09
1.33E-09
8.82E-08
2.85E-07
3.67E-10
*OLINDA/EXM v1.0.
Note: GI, gastrointestinal; SI, small intestine; ULI, upper large intestine; LLI, lower large intestine; Bold italicized correspond to values in MIRD example problem. E, exponential
(e.g., 4.6E-05 4.6 1025)
5.66E-08
8.11E-09
Thymus
Thyroid
4.58E-07
1.54E-09
Spleen
Testes
626
Whole Body Dose and Effective Dose

Dose factor tables always include entries for the whole body of the reference adult,
child, or pregnant woman. As is the case for individual organs, all energy absorbed
from all emissions are summed and then divided by the mass, in this case, of the
whole body. Researchers and regulators usually note this quantity, but in nuclear
medicine, it is of questionable utility. If the whole body is fairly uniformly irradiated,
this quantity may be useful in risk assessment, but this is rarely the case with radiopharmaceuticals, as they are specifically designed to target specific tissues. Averaging
all of the energy absorbed in particular organs over the mass of the whole body may
give a misleading result. As an example, imagine that we have I-131 activity only in
the thyroid of an adult male. Lets say there were 100 MBq-hr in the thyroid, whose
mass is about 20 g. The dose to the thyroid is
mGy
3600s
= 570mGy
D = 100MBq-hr
1.59 10-3
hr
MBq - s
One can use the dose factor for thyroid to total body and calculate a total body
dose of 0.27 mGy but it is of very little value. The dose to the thyroid, on the other
hand, can be useful in evaluating the potential risk by using the ICRP 103 methodology to calculate an effective dose (see Chapter 3) or to calculate the relative or
absolute thyroid cancer risk from the dose based on organ-specific risk estimates
(see discussion of risk calculations in Chapter 20). However, in the RADAR or MIRD
dosimetry systems, the whole body dose is calculated by taking all of the beta
energy that was absorbed in the 20 g of thyroid and averaging it over 73,000 g of
body tissue (mass of the reference person) and adding the photon energy that was
fairly uniformly received by tissues outside of the thyroid. There is little one can do
with a dose calculated in this manner that would have any relevence to the assessment of potential health risks.
Recall (from Chapter 3) that in the effective dose model, tissue weighting factors
(wT) are used to account for the differences among individual organs and tissues to harm
(i.e., detriment) from radiation-induced cancer, genetic, and other stochastic effects.
The wT values were chosen on the basis of age and gender averaged estimates of
relative radiation induced detriment for each of organs and tissues listed. To reflect
the combined detriment from stochastic effects due to the equivalent doses in all the
organs and tissues of the body, the equivalent dose in each organ and tissue is multiplied by its corresponding wT, and the results are added to give the effective dose.
E = HT wT
T
If one were to recalculate the effective dose for the Tc-99m-sulfur colloid dosimetry
example illustrated above, this time including all the source and target organs, more
realistic kinetics, and a more typical administered activity for an adult liver/spleen
scan of 200 MBq, the effective dose would be approximately 1.9 mSv. This value
could be used to compare the relative risk of the Tc-99m-sulfur colloid liver scan
to an alternative diagnostic imaging procedure using ionizing radiation such as an
abdominal CT scan for which the effective dose (8 mSv), and thus the risk, may be
approximately four times higher.
It is important to remember that the effective dose concept was developed for
the purpose of radiation protection where the reference population for which the wT
values were developed included a distribution of both genders and all ages. Effective dose was not intended, nor is it correct to assume that it can be used, to provide
individual risk estimates to a particular patient from medical imaging procedures.

This is a consequence of the fact that the magnitude of risk for stochastic effects are
dependent on variables beyond just organ/tissue doses such as age at time of exposure and gender. For example, as discussed in Chapter 20, radiation exposure in
children carriers a greater cancer risk than the same dose in adult patients. Thus, calculating risk on the basis of effective doses for a pediatric patient population would
likely underestimate the actual cancer risk and such calculations are inappropriate
anytime there are significant dissimilarities between the age and gender of a particular patient or patient population and the ICRP reference population. The effective
dose methodology was not developed to assess radiation-induced health risks in
patients receiving radiopharmaceutical therapy, even if the patient populations are
similar to the age and gender distributions of the ICRP reference population.
Accuracy of Dose Calculations

Although the methods shown here provide reasonable estimates of organ doses, the
typical application of this technique usually includes several significant assumptions,
limitations, and simplifications that, taken together, could result in significant differences between the true and calculated doses. These include the following:
1. The radioactivity is assumed to be uniformly distributed in each source organ.
This is rarely the case and, in fact, significant pathology (e.g., cirrhosis of the
liver) or characteristics of the radiopharmaceutical may result in a highly nonuniform activity distribution.
2. The dose factors derived from early computational models were based on organ
sizes and geometries that were idealized into simplified shapes to reduce the computational complexity. Significant improvements have been made over the last
few decades and more detailed and anatomically correct anthropomorphic phantoms are now available.
3. Each organ is assumed to be homogeneous in density and composition.
4. Even the newer more detailed reference adult, adolescent, and child phantoms
are just approximations of the physical dimensions of any given individual.
5. Although the radiobiologic effect of the dose occurs at the molecular level, the
energy deposition is averaged over the entire mass of the target organs and therefore does not reflect the actual microdosimetry on a molecular or cellular level.
6. Dose contributions from bremsstrahlung and other minor radiation sources are
ignored.
7. With a few exceptions, low-energy photons and all particulate radiations are
assumed to be absorbed locally (i.e., nonpenetrating).
Taken together, radiopharmaceutical dose estimate may be different (higher or lower)
by a factor of two or more than the actual dose in any given individual. This is particularly true in patients with significant disease states that alter the kinetics of the normal
radiopharmaceutical distribution. However, for patients who have had individualized
dosimetry performed prior to receiving radionuclide therapy, the total uncertainty in
an individual dose estimate can be reduced to a value of perhaps 10% to 20% (Stabin
2008). In addition, the time-activity curves used for each initial organ dose estimate
during the developmental phase of a radiopharmaceutical is usually based on laboratory animal data. This information is only slowly adjusted by data aquired in human
subjects and quantitative evaluation of biodistributions and kinetics. The FDA does
not currently require manufacturers to update their package inserts as better radiopharmaceutical dosimetry becomes available. Therefore, organ doses listed in package
inserts (especially those of older agents) are often not the best source of dosimetry
information (Appendix H, Fig. H-3). The ICRP, utilizing similar methodology, reference
628
data on biokinetics, and new ORNL phantoms discussed above, have compiled the
most comprehensive collection of dose estimates for radiopharmaceuticals in their
publications 53, 80, and 106 (ICRP 1987, ICRP 1998, ICRP 2009).
In addition to the development of child and fetal dosimetry models, advances
in the use of radioimmunotherapeutic pharmaceuticals have increased the need for
patient-specific dosimetry that takes advantage of individual kinetics and anatomic
information. A good overview of these issues and models can be found in several
recent reviews (see suggested reading).
For the most commonly used diagnostic and therapeutic radiopharmaceutical
agents, Appendix F-2 summarizes the typically administered adult dose, the organ
receiving the highest radiation dose and its dose, the gonadal dose, and the adult
effective dose. For most of these same radiopharmaceuticals, Appendix F-3A provides a table of effective doses per unit activity administered in 15-, 10-, 5-, and
1-year-old patients, Appendix F-3B provides the North American consensus guidelines for administered radiopharmaceutical activities in children and adolescents,
Appendix F4-A provides a table of absorbed doses to the embryo or fetus at early, 3,
6, and 9 months gestation per unit activity of commonly used radiopharmaceuticals
administered to the mother, Appendix F4-B provides a table of effective doses to the
newborn and infant per unit activity of specific radiopharmaceuticals administered
from the mothers breast milk and Appendix F4-C provides a table of breast dose
from radiopharmaceuticals excreted in breast milk.
16.4 Regulatory Issues
Investigational Radiopharmaceuticals
All pharmaceuticals for human use, whether radioactive or not, are regulated by the
U.S. Food and Drug Administration (FDA). A request to evaluate a new radiopharmaceutical for human use is submitted to the FDA in an application called a Notice
of Claimed Investigational Exemption for a New Drug (IND). The IND can be sponsored by either an individual physician or a radiopharmaceutical manufacturer who
will work with a group of clinical investigators to collect the necessary data. The IND
application includes the names and credentials of the investigators, the clinical protocol, details of the research project, details of the manufacturing of the drug, and animal
toxicology data. The clinical investigation of the new radio pharmaceutical occurs in
three stages. Phase I focuses on a limited number of patients and is designed to provide
information on the pharmacologic distribution, metabolism, dosimetry, toxicity, optimal dose schedule, and adverse reactions. Phase II studies include a limited number of
patients with specific diseases to begin the assessment of the drugs efficacy, refine the
dose schedule, and collect more information on safety. Phase III clinical trials involve
a much larger number of patients (and are typically conducted by several institutions)
to provide more extensive (i.e., statistically significant) information on efficacy, safety,
and dose administration. To obtain approval to market a new radiopharmaceutical,
a New Drug Application (NDA) must be submitted to and approved by the FDA.
Approval of a new radiopharmaceutical typically requires 5 to 10 years from laboratory work to NDA. The package insert of an approved radiopharmaceutical describes
the intended purpose of the radiopharmaceutical, the suggested dose, dosimetry,
adverse reactions, clinical pharmacology, and contraindications.
Any research involving human subjects conducted, supported, or otherwise regulated by a federal department or agency must be conducted in accordance with the
Federal Policy for the Protection of Human Subjects, (Federal Register, 1991); this
policy is codified in the regulations of 15 federal departments and agencies. This policy requires that all research involving human subjects be reviewed and approved by
an institutional review board (IRB) and that informed consent be obtained from each
research subject. Most academic medical institutions have IRBs. An IRB comprises
clinical, scientific, legal, and other experts and must include at least one member
who is not otherwise affiliated with the institution. Informed consent must be sought
in a manner that minimizes the possibility of coercion and provides the subject sufficient opportunity to decide whether or not to participate. The information presented
must be in language understandable to the subject. It must include a statement that
the study involves research, the purposes of the research, a description of the procedures, and identification of any procedures that are experimental; a description of
any reasonably foreseeable risks or discomforts; a description of any likely benefits
to the subject or to others; a disclosure of alternative treatments; and a statement that
participation is voluntary, that refusal will involve no penalty or loss of benefits, and
that the subject may discontinue participation at any time.
Byproduct Material, Authorized Users, Written Directives,

and Medical Events
Medical use of Byproduct Material
Although the production of radiopharmaceuticals is regulated by the FDA, the medical use of radioactive material is regulated under the terms of a license issued to a
specific legal entity (such as a clinic or hospital which is the licensee) by the U.S.
Nuclear Regulatory Commission (NRC) or, a comparable state agency (i.e., an agreement state, which is discussed further in Chapter 21). The NRCs regulations apply
to the use of byproduct material. Until recently, the regulatory definition of byproduct material included radionuclides that were the byproducts of nuclear fission or
nuclear activation but excluded others such as accelerator-produced radionuclides.
The current NRCs definition of byproduct material, however, has been broadened to
include virtually all radioactive material used in medicine. The regulations regarding
the medical use of radioactive material are contained in Title 10, Part 35, of the Code
of Federal Regulations (10 CFR 35).
Authorized User
An authorized user (AU), in the context of the practice of nuclear medicine, is a physician who is responsible for the medical use of radioactive material and is designated
by name on a license for the medical use of radioactive material or is approved by
the radiation safety committee of a medical institution whose license authorizes such
actions. Such a physician may be certified by a medical specialty board, such as the
American Board of Radiology, whose certification process includes all of the requirements identified in Part 35 for the medical use of unsealed sources of radioactive
materials for diagnosis and therapy. Alternatively a physician can apply to the NRC
or comparable state agency for AU status by providing documentation of the specific
education, training and experience requirements contained in 10 CFR 35.
Written Directive
The NRC requires that, before the administration of a dosage of I-131 sodium iodide
greater than 1.11 MBq (30 mCi) or any therapeutic dosage of unsealed byproduct
material, a written directive must be signed and dated by an authorized user. The
630
written directive must contain the patient or human research subjects name and
must describe the radioactive drug, the activity, and (for radionuclides other than
I-131) the route of administration. In addition, the NRC requires the implementation
of written procedures to provide, for each administration requiring a written directive, high confidence that the patient or human research subjects identity is verified
before the administration and that each administration is performed in accordance
with the written directive.
Medical Events
The NRC defines certain errors in the administration of radiopharmaceuticals as
medical events, and requires specific actions to be taken within specified time periods following the recognition of the error. The initial report to the NRC, (or, in an
agreement state, the comparable state agency) must be made by telephone no later
than the next calendar day after the discovery of the event and must be followed
by a written report within 15 days. This report must include specific information
such as a description of the incident, the cause of the medical event, the effect (if
any) on the individual or individuals involved, and proposed corrective actions.
The referring physician must be notified of the medical event, and the patient must
also be notified, unless the referring physician states that he or she will inform the
patient or that, based on medical judgment, notification of the patient would be
harmful. Additional details regarding these reporting requirements can be found
in 10 CFR 35.
The NRC defines a medical event as:
A.The administration of NRC-licensed radioactive materials that results in one of
the following conditions (1, 2 or 3 below) unless its occurrence was as the direct
result of patient intervention (e.g., an I-131 therapy patient takes only one-half
of the prescribed treatment and then refuses to take the balance of the prescribed
dosage):
1.A dose that differs from the prescribed dose or dose that would have resulted
from the prescribed dosage (i.e., administered activity) by more than 0.05 Sv (5
rem) effective dose equivalent, 0.5 Sv (50 rem) to an organ or tissue, or 0.5 Sv
(50 rem) shallow dose equivalent to the skin; and one of the following conditions (i or ii) has also occurred.
(i)The total dose delivered differs from the prescribed dose by 20% or more;
(ii)The total dosage delivered differs from the prescribed dosage by 20%
or more or falls outside the prescribed dosage range. Falling outside
the prescribed dosage range means the administration of activity that is
greater or less than a predetermined range of activity for a given procedure that has been established by the licensee.
2. A dose that exceeds 0.05 Sv (5 rem) effective dose equivalent, 0.5 Sv
(50 rem) to an organ or tissue, or 0.5 Sv (50 rem) shallow dose equivalent to
the skin from any of the following
(i)An administration of a wrong radioactive drug containing byproduct material;
(ii)An administration of a radioactive drug containing byproduct material
through the wrong route of administration;
(iii)An administration of a dose or dosage to the wrong individual or human
research subject.
3. A dose to the skin or an organ or tissue other than the treatment site that
exceeds by 0.5 Sv (50 rem) and 50% or more of the dose expected from the
administration defined in the written directive.
B.Any event resulting from intervention of a patient or human research subject
in which the administration of byproduct material or radiation from byproduct
material results or will result in unintended permanent functional damage to an
organ or a physiological system, as determined by a physician.
This definition of a medical event was summarized from NRC regulations. It
applies only to the use of unsealed byproduct material and omits the definition
of a medical event involving the use of sealed sources of byproduct material to
treat patients (e.g., conventional brachytherapy treatment of prostate cancer with
I-125 seeds by radiation oncologists). The complete regulations regarding written
directives, authorized users and medical events can be found in 10 CFR 35. State
regulatory requirements should be consulted, because they may differ from federal
regulations.
SUGGESTED READING
Dillehay G, Ellerbroek Balon H, et al. Practice guideline for the performance of therapy with unsealed
radiopharmaceutical sources. Int J Radiat Oncol Biol Phys 2006;64(5,1):12991307.
Hung JC, Ponto JA, Hammes RJ. Radiopharmaceutical-related pitfalls and artifacts. Semin Nucl Med
1996;26:208255.
Macey DJ, Williams LE, Breitz HB, et al. AAPM report no.71, A primer for radioimmunotherapy and
radionuclide therapy. Madison, WI: Medical Physics Publishing; 2001.
Medley CM, Vivian GC. Radionuclide developments. Br J Radiol 1997;70:133144.
Ponto JA. The AAPM/RSNA physics tutorial for residents: radiopharmaceuticals. Radiographics
1998;18:13951404.
Rhodes BA, Hladik WB, Norenberg JP. Clinical radiopharmacy: principles and practices. Semin Nud Med
1996;26:7784.
Saha GB. Fundamentals of nuclear pharmacy. 5th ed. New York, NY: Springer-Verlag, 2004.
Sampson CB, ed. Textbook of radiopharmacy: theory and practice. 2nd ed. New York, NY: Gordon and
Breach Publishers, 1995.
Loevinger R, Budinger T, Watson E. MIRD primer for absorbed dose calculations. New York, NY: Society
of Nuclear Medicine, 1991.
Stabin M. Nuclear medicine dosimetry. Phys Med Biol 2006:51.
Toohey RE, Stabin MG, Watson EE. The AAPM/RSNA physics tutorial for residents. Internal radiation
dosimetry: principles and applications. Radiographics 2000;20:533546.
Zanzonico PB. Internal radionuclide radiation dosimetry: a review of basic concepts and recent
developments. J Nuclear Med 2000;41:297308.
SELECTED REFERENCES
Bolch WE, Eckerman KF, Sgouros G, Thomas SR. MIRD Pamphlet No. 21: A Generalized Schema
for Radiopharmaceutical DosimetryStandardization of nomenclature. J Nucl Med 2009;50:
477484.
Cristy M, Eckerman K. Specific absorbed fractions of energy at various ages from internal photons sources.
ORNL/TM-8381 V1-V7. Oak Ridge, TN: Oak Ridge National Laboratory, 1987.
Federal Register 1991: Federal Register. Federal Policy for the Protection of Human Subjects. June 18, 1991:
2800328032.
ICRP 1987: Publication 53: Biokinetics and Dosimetry: General Considerations. Ann ICRP 1987;
18(14).
ICRP 1998: ICRP Publication 80: Radiation Dose to Patients from Radiopharmaceuticals: Addendum 2 to
ICRP Publication 53. Ann ICRP 1998;28(3).
ICRP 2008: ICRP Publication 106: Radiation Dose to Patients from Radiopharmaceuticals A Third
Addendum to ICRP Publication 53 ICRP. Ann ICRP. 2008 38(12).
Robbins RJ. Chromatography of Technetium-99m Radiopharmaceuticals A Practical Guide. Society of Nuclear
Medicine, Reston, VA, 1985.
632

Stabin M, Watson E, Cristy M, Ryman J, Eckerman K, Davis J, Marshall D, Gehlen K. mathematical models
and specific absorbed fractions of photon energy in the nonpregnant adult female and at the end of
each trimester of pregnancy. ORNL Report ORNL/TM 1995;12907.
Stabin MG. MIRDOSE: personal computer software for internal dose assessment in nuclear medicine.
J Nucl Med 1996;37(3):538546.
Stabin MG. Uncertainties in internal dose calculations for radiopharmaceuticals. J Nucl Med 2008;49:
853860.
Stabin MG, Siegel JA. Physical models and dose factors for use in internal dose assessment. Health Physics
2003;85(3):294310.
Stabin MG, Sparks RB, Crowe E. OLINDA/EXM: the second-generation personal computer software for
internal dose assessment in nuclear medicine. J Nucl Med 2005;46.
Yalow, RS, Berson, SA. Immunoassay of endogenous plasma insulin in main. J clin invest 1960;39:
11571175.
Chapter
17
Radiation Detection
andMeasurement
The detection and measurement of ionizing radiation are the basis for the majority
of diagnostic imaging. In this chapter, the basic concepts of radiation detection and
measurement are introduced, followed by a discussion of the characteristics of specific types of detectors. The electronic systems used for pulse height spectroscopy
and the use of sodium iodide (NaI) scintillators to perform gamma-ray spectroscopy
are described, followed by a discussion of detector applications. The use of radiation
detectors in imaging devices is covered in other chapters.
All detectors of ionizing radiation require the interaction of the radiation with
matter. Ionizing radiation deposits energy in matter by ionization and excitation.
Ionization is the removal of electrons from atoms or molecules. (An atom or molecule
stripped of an electron has a net positive charge and is called a cation. In many gases,
the free electrons become attached to uncharged atoms or molecules, forming negatively charged anions. An ion pair consists of a cation and its associated free electron
or anion.) Excitation is the elevation of electrons to excited states in atoms, molecules,
or a crystal. Excitation and ionization may produce chemical changes or the emission
of visible light or ultraviolet (UV) radiation. Most energy deposited by ionizing radiation is ultimately converted into thermal energy.
The amount of energy deposited in matter by a single interaction is very small. For
example, a 140-keV gamma ray deposits 2.24 1014 joules if completely absorbed.
To raise the temperature of 1 g of water by 1C (i.e., 1 calorie) would require the
complete absorption of 187 trillion (187 1012) of these photons. For this reason,
most radiation detectors provide signal amplification. In detectors that produce an
electrical signal, the amplification is electronic. In photographic film, the amplification is achieved chemically.
17.1 Types of Detectors and Basic Principles

Radiation detectors may be classified by their detection method. A gas-filled detector
consists of a volume of gas between two electrodes. Ions produced in the gas by the
radiation are collected by the electrodes, resulting in an electrical signal.
The interaction of ionizing radiation with certain materials produces ultraviolet
radiation and/or visible light. These materials are called scintillators. They are commonly attached to or incorporated in devices that convert the UV radiation and light
into an electrical signal. For other applications, photographic film is used to record
the light emitted by the scintillators. Many years ago, in physics research and medical
fluoroscopy, the light from scintillators was viewed directly with dark-adapted eyes.
Semiconductor detectors are especially pure crystals of silicon, germanium, or other
semiconductor materials to which trace amounts of impurity atoms have been added
so that they act as diodes. A diode is an electronic device with two terminals that
permits a large electrical current to flow when a voltage is applied in one direction,
633
634
but very little current when the voltage is applied in the opposite direction. When
used to detect radiation, a voltage is applied in the direction in which little current
flows. When an interaction occurs in the crystal, electrons are raised to an excited
state, allowing a momentary pulse of electrical current to flow through the device.
Detectors may also be classified by the type of information produced. Detectors, such as Geiger-Mueller (GM) detectors, that indicate the number of interactions
occurring in the detector are called counters. Detectors that yield information about
the energy distribution of the incident radiation, such as NaI scintillation detectors,
are called spectrometers. Detectors that indicate the net amount of energy deposited
in the detector by multiple interactions are called dosimeters.
Pulse and Current Modes of Operation

Many radiation detectors produce an electrical signal after each interaction of a particle or photon. The signal generated by the detector passes through a series of electronic circuits, each of which performs a function such as signal amplification, signal
processing, or data storage. A detector and its associated electronic circuitry form a
detection system. There are two fundamental ways that the circuitry may process the
signalpulse mode and current mode. In pulse mode, the signal from each interaction is processed individually. In current mode, the electrical signals from individual
interactions are averaged together, forming a net current signal.
There are advantages and disadvantages to each method of handling the signal.
GM detectors are operated in pulse mode, whereas most ionization chambers, including ion chamber survey meters and the dose calibrators used in nuclear medicine,
are operated in current mode. Scintillation detectors are operated in pulse mode in
nuclear medicine applications, but in current mode in direct digital radiography,
fluoroscopy, and x-ray computed tomography (CT).
In this chapter, the term interaction typically refers to the interaction of a single
photon or charged particle, such as the interaction of a gamma ray by the photoelectric
effect or Compton scattering. The term event may refer to a single interaction, or it
may refer to something more complex, such as two nearly simultaneous interactions
in a detector. In instruments which process the signals from individual interactions or
events in pulse mode, an interaction or event that is registered is referred to as a count.
Effect of Interaction Rate on Detectors Operated in Pulse Mode

The main problem with using a radiation detector or detection system in pulse mode
is that two interactions must be separated by a finite amount of time if they are to
produce distinct signals. This interval is called the dead time of the system. If a second
interaction occurs during this time interval, its signal will be lost; furthermore, if it is
close enough in time to the first interaction, it may even distort the signal from the
first interaction. The fraction of counts lost from dead-time effects is smallest at low
interaction rates and increases with increasing interaction rate.
The dead time of a detection system is largely determined by the component in the
series with the longest dead time. For example, the detector usually has the longest dead
time in GM counter systems, whereas in multichannel analyzer (MCA) systems (see later
discussion), the analog-to-digital converter (ADC) generally has the longest dead time.
The dead times of different types of systems vary widely. GM counters have dead
times ranging from tens to hundreds of microseconds, whereas most other systems
have dead times of less than a few microseconds. It is important to know the countrate behavior of a detection system; if a detection system is operated at too high an
interaction rate, an artificially low count rate will be obtained.
There are two mathematical models describing the behavior of detector systems
operated in pulse modeparalyzable and nonparalyzable. Although these models
Chapter 17 Radiation Detection andMeasurement
10,000
635
Ideal
Non-paralyzable
Count Rate
8000
6000
4000
2000
Paralyzable
10,000
20,000
30,000
Interaction Rate
40,000
50,000
FIGURE 17-1 Effect of interaction rate on measured count rate of paralyzable and nonparalyzable detectors. The ideal line represents the response of a hypothetical detector that does not suffer from dead-time
count losses (i.e., the count rate is equal to the interaction rate). Note that the y-axis scale is expanded with
respect to that of the x-axis; the ideal line would be at a 45-degree angle if the scales were equal.
are simplifications of the behavior of real detection systems, real systems may behave
like one or the other model. In a paralyzable system, an interaction that occurs during
the dead time after a previous interaction extends the dead time; in a nonparalyzable
system, it does not. Figure 17-1 shows the count rates of paralyzable and nonparalyzable detector systems as a function of the rate of interactions in the detector. At very
high interaction rates, a paralyzable system will be unable to detect any interactions
after the first, because subsequent interactions will extend the dead time, causing the
system to indicate a count rate of zero!
Current Mode Operation

When a detector is operated in current mode, all information regarding individual
interactions is lost. For example, neither the interaction rate nor the energies deposited by individual interactions can be determined. However, if the amount of electrical charge collected from each interaction is proportional to the energy deposited by
that interaction, then the net electrical current is proportional to the dose rate in the
detector material. Detectors subject to very high interaction rates are often operated
in current mode to avoid dead-time information losses. Image-intensifier tubes and
flat panel image receptors in fluoroscopy, detectors in x-ray CT machines, direct digital radiographic image receptors, ion chambers used in phototimed radiography, and
most nuclear medicine dose calibrators are operated in current mode.
Spectroscopy
The term spectroscopy, literally the viewing of a spectrum, is commonly used to refer
to measurements of the energy distributions of radiation fields, and a spectrometer is
a detection system that yields information about the energy distribution of the incident radiation. Most spectrometers are operated in pulse mode, and the amplitude of
each pulse is proportional to the energy deposited in the detector by the interaction
causing that pulse. The energy deposited by an interaction, however, is not always the
total energy of the incident particle or photon. For example, a gamma ray may interact
with the detector by Compton scattering, with the scattered photon escaping the
detector. In this case, the deposited energy is the difference between the energies of
the incident and scattered photons. A pulse height spectrum is usually depicted as a
100
662 keV
32 keV
0
100 200 300 400 500 600 700

Energy (keV)
32 keV
Number of Interactions
Relative Number of Photons
636
662 keV
100
200
300 400 500

Energy (keV)
600
700
FIGURE 17-2 Energy spectrum of cesium-137 (left) and resultant pulse height spectrum from a detector
(right).
graph of the number of interactions depositing a particular amount of energy in the

spectrometer as a function of energy (Fig. 17-2). Because the energy deposited by an
interaction may be less than the total energy of the incident particle or photon and
also because of random variations in the detection process, the pulse height spectrum
produced by a spectrometer is not identical to the actual energy spectrum of the incident
radiation. The energy resolution of a spectrometer is a measure of its ability to differentiate between particles or photons of different energies. Pulse height spectroscopy
is discussed later in this chapter.
Detection Efficiency
The efficiency (sensitivity) of a detector is a measure of its ability to detect radiation.
The efficiency of a detection system operated in pulse mode is defined as the probability that a particle or photon emitted by a source will be detected. It is measured
by placing a source of radiation in the vicinity of the detector and dividing the number of particles or photons detected by the number emitted:

Efficiency
Number detected
Number emitted
[17-1]
This equation can be written as follows:

Efficiency
Number reaching detector

Number detected
Number emitted
Number reaching detector
Therefore, the detection efficiency is the product of two terms, the geometric
efficiency and the intrinsic efficiency:

Efficiency Geometric efficiency Intrinsic efficiency
[17-2]
where the geometric efficiency of a detector is the fraction of emitted particles or photons that reach the detector and the intrinsic efficiency is the fraction of those particles
or photons reaching the detector that are detected. Because the total, geometric, and
intrinsic efficiencies are all probabilities, each ranges from 0 to 1.
The geometric efficiency is determined by the geometric relationship between the
source and the detector (Fig. 17-3). It increases as the source is moved toward the
detector and approaches 0.5 when a point source is placed against a flat surface of
the detector, because in that position one half of the photons or particles are emitted
into the detector. For a source inside a well-type detector, the geometric efficiency
approaches 1, because most of the particles or photons are intercepted by the detector.
(A well-type detector is a detector containing a cavity for the insertion of samples.)
637
source
source
source
detector
detector
detector
FIGURE 17-3 Geometric efficiency. With a source far from the detector (left), the geometric efficiency is
less than 50%. With a source against the detector (center), the geometric efficiency is approximately 50%.
With a source in a well detector (right), the geometric efficiency is greater than 50%.
The intrinsic efficiency of a detector in detecting photons, also called the quantum
detection efficiency (QDE), is determined by the energy of the photons and the atomic
number, density, and thickness of the detector. If a parallel beam of monoenergetic
photons is incident upon a detector of uniform thickness, the intrinsic efficiency of
the detector is given by the following equation:

Intrinsic efficiency 1 ex 1 e( /)x
[17-3]
where m is the linear attenuation coefficient of the detector material, is the density of the material, m/ is the mass attenuation coefficient of the material, and x is
the thickness of the detector. This equation shows that the intrinsic efficiency for
detecting x-rays and gamma rays increases with the thickness of the detector and
the density and the mass attenuation coefficient of the detector material. The mass
attenuation coefficient increases with the atomic number of the material and, within
the range of photon energies used in diagnostic imaging, decreases with increasing
photon energy, with the exception of absorption edges (Chapter 3).
17.2 Gas-Filled Detectors
Basic Principles
A gas-filled detector (Fig. 17-4) consists of a volume of gas between two electrodes,
with an electric potential difference (voltage) applied between the electrodes. Ionizing radiation forms ion pairs in the gas. The positive ions (cations) are attracted
to the negative electrode (cathode), and the electrons or anions are attracted to the
positive electrode (anode). In most detectors, the cathode is the wall of the container
that holds the gas or a conductive coating on the inside of the wall, and the anode is
a wire inside the container. After reaching the anode, the electrons travel through the
circuit to the cathode, where they recombine with the cations. This electrical current
can be measured with a sensitive ammeter or other electrical circuitry.
There are three types of gas-filled detectors in common useionization chambers, proportional counters, and GM counters. The type of detector is determined
638

meter
+ + + + +
anode (+)
incident
charged
particle
cathode (-)
battery
or power supply
FIGURE 17-4 Gas-filled detector. A charged particle, such as a beta particle, is shown entering the detector
from outside and creating ion pairs in the gas inside the detector. This can occur only if the detector has a sufficiently thin wall. When a thick-wall gas-filled detector is used to detect x-rays and gamma rays, the charged
particles causing the ionization are mostly electrons generated by Compton and photoelectric interactions of
the incident x-rays or gamma rays in the detector wall or in the gas in the detector.
primarily by the voltage applied between the two electrodes. In an ionization chamber, the two electrodes can have almost any configuration: they may be two parallel
plates, two concentric cylinders, or a wire within a cylinder. In proportional counters
and GM counters, the anode must be a thin wire. Figure 17-5 shows the amount
of electrical charge collected after a single interaction as a function of the electrical
potential difference (voltage) applied between the two electrodes.
Ionizing radiation produces ion pairs in the gas of the detector. If no voltage is
applied between the electrodes, no current flows through the circuit because there
is no electric field to attract the charged particles to the electrodes; the ion pairs
merely recombine in the gas. When a small voltage is applied, some of the cations
1012
Limited
Proportionality
Geiger-Mueller
Region
1010
Number of Ion Pairs Collected
Continuous
Discharge
Recombination
108
106
104
Ionization Chamber
Region
Proportional Region
1 MeV beta
particles
100 keV beta

particles
102
100
Applied Voltage
FIGURE 17-5 Amount of electrical charge collected after a single interaction as a function of the electrical potential difference (voltage) applied between the two electrodes of a gas-filled detector. The lower curve shows the
charge collected when a 100-keV electron interacts, and the upper curve shows the result from a 1-MeV electron.
639
are attracted to the cathode and some of the electrons or anions are attracted to the
anode before they can recombine. As the voltage is increased, more ions are collected
and fewer recombine. This region, in which the current increases as the voltage is
raised, is called the recombination region of the curve.
As the voltage is increased further, a plateau is reached in the curve. In this region,
called the ionization chamber region, the applied electric field is sufficiently strong to
collect almost all ion pairs; additional increases in the applied voltage do not significantly increase the current. Ionization chambers are operated in this region.
Beyond the ionization region, the collected current again increases as the applied
voltage is raised. In this region, called the proportional region, electrons approaching
the anode are accelerated to such high kinetic energies that they cause additional ionization. This phenomenon, called gas multiplication, amplifies the collected current;
the amount of amplification increases as the applied voltage is raised.
At any voltage through the ionization chamber region and the proportional
region, the amount of electrical charge collected from each interaction is proportional
to the amount of energy deposited in the gas of the detector by the interaction. For
example, the amount of charge collected after an interaction depositing 100 keV is
one tenth of that collected from an interaction depositing 1 MeV.
Beyond the proportional region is a region in which the amount of charge collected from each event is the same, regardless of the amount of energy deposited by
the interaction. In this region, called the Geiger-Mueller region (GM region), the gas
multiplication spreads the entire length of the anode. The size of a pulse in the GM
region tells us nothing about the energy deposited in the detector by the interaction
causing the pulse. Gas-filled detectors cannot be operated at voltages beyond the GM
region because they continuously discharge.
Ionization Chambers (Ion Chambers)

Because gas multiplication does not occur at the relatively low voltages applied to ionization chambers, the amount of electrical charge collected from a single interaction is very
small and would require huge amplification to be detected. For this reason, ionization
chambers are seldom used in pulse mode. The advantage to operating them in current
mode is the almost complete freedom from dead-time effects, even in very intense radiation fields. In addition, as shown in Figure 17-5, the voltage applied to an ion chamber
can vary significantly without appreciably changing the amount of charge collected.
Almost any gas can be used to fill the chamber. If the gas is air and the walls of the
chamber are of a material whose effective atomic number is similar to air, the amount of
current produced is proportional to the exposure rate (exposure is the amount of electrical charge produced per mass of air). Air-filled ion chambers are used in portable survey
meters and can accurately indicate exposure rates from less than 1 mR/h to tens or hundreds of roentgens per hour (Fig. 17-6). Air-filled ion chambers are also used for performing quality-assurance testing of diagnostic and therapeutic x-ray machines, and they
are the detectors in most x-ray machine phototimers. Measurements using an air-filled
ion chamber that is open to the atmosphere are affected by the density of the air in the
chamber, which is determined by ambient air pressure and temperature. Measurements
using such chambers that require great accuracy must be corrected for these factors.
In very intense radiation fields, there can be signal loss due to recombination
of ions before they are collected at the electrodes, causing the current from an ion
chamber to deviate from proportionality to the intensity of the radiation. An ion
chamber intended for use in such fields may have a small gas volume, a low gas density, and/or a high applied voltage to reduce this effect.
640
FIGURE 17-6 Portable air-filled ionization chamber survey meter (A). This particular instrument measures exposure rates ranging from about 0.1 mR/h to 50 R/h. The exposure rate is indicated by the position of the red needle
on the scale. The scale is selected using the range knob located below the scale (B). In this case, the needle is pointing to a value of 0.6 on the scale, and the range selector is set at 50 mR/h. Thus, the exposure rate being shown
is 6 mR/h. The interior of the instrument is shown (C) and the ion chamber, covered with a thin Mylar membrane,
is easily seen. On the bottom of the meter case (D) is a slide (E) that can cover or expose the thin Mylar window
of the ion chamber. This slide should be opened when measuring low-energy x-ray and gamma-ray radiation. The
slide can also be used to determine if there is a significant beta radiation component in the radiation being measured. If there is no substantial change in the measured exposure rate with the slide open (where beta radiation
can penetrate the thin membrane and enter the ion chamber) or closed (where the ion chamber is shielded from
beta radiation), the radiation can be considered to be comprised primarily of x-rays or gamma rays.
Gas-filled detectors tend to have low intrinsic efficiencies for detecting x-rays
and gamma rays because of the low densities of gases and the low atomic numbers of
most common gases. The sensitivity of ion chambers to x-rays and gamma rays can
be enhanced by filling them with a gas that has a high atomic number, such as argon
(Z 18) or xenon (Z 54), and pressurizing the gas to increase its density. Welltype ion chambers called dose calibrators are used in nuclear medicine to assay the
activities of dosages of radiopharmaceuticals to be administered to patients; many are
filled with pressurized argon. Xenon-filled pressurized ion chambers were formerly
used as detectors in some CT machines.
Air filled ion chambers are commonly used to measure the related quantities air
kerma and exposure rate. These quantities were defined in Chapter 3. Air kerma is the
initial kinetic energy transferred to charged particles, in this case electrons liberated
in air by the radiation, per mass air and exposure is the amount of electrical charge
created in air by ionization caused by these electrons, per mass air. There is a problem measuring the ionization in the small volume of air in an ionization chamber of
reasonable size. The energetic electrons released by interactions in the air have long
ranges in air and many of them would escape the air in the chamber and cause much
641
of their ionization elsewhere. This problem can be partially solved by building the
ion chamber with thick walls of a material whose effective atomic number is similar
to that of air. In this case, the number of electrons escaping the volume of air is
approximately matched by a similar number of electrons released in the chamber
wall entering the air in the ion chamber. This situation, if achieved, is called electronic equilibrium. For this reason, most ion chambers for measuring exposure or air
kerma have thick air-equivalent walls, or are equipped with removable air-equivalent
buildup caps to establish electronic equilibrium. The thickness of material needed to
establish electronic equilibrium increases with the energy of the x- or gamma rays.
However, thick walls or buildup caps may significantly attenuate low energy x- and
gamma rays. Many ion chamber survey meters have windows that may be opened
in the thick material around the ion chamber to permit more accurate measurement
of low energy x- and gamma rays.Electronic equilibrium, also called charged particle
equilibrium, is discussed in detail in more advanced texts (Attix, 1986; Knoll, 2010).
Proportional Counters
Unlike ion chambers, which can function with almost any gas, including air, a
proportional counter must contain a gas with low electron affinity, so that few free
electrons become attached to gas molecules. Because gas multiplication can produce a
charge-per-interaction that is hundreds or thousands of times larger than that produced
by an ion chamber, proportional counters can be operated in pulse mode as counters
or spectrometers. They are commonly used in standards laboratories, in health physics
laboratories, and for physics research. They are seldom used in m
edical centers.
Multiwire proportional counters, which indicate the position of an interaction
in the detector, have been studied for use in nuclear medicine imaging devices. They
have not achieved acceptance because of their low efficiencies for detecting x-rays
and gamma rays from the radionuclides commonly used in nuclear m
edicine.
Geiger-Mueller Counters
GM counters must also contain gases with specific properties, discussed in more
advanced texts. Because gas multiplication produces billions of ion pairs after an
interaction, the signal from a GM detector requires little additional amplification. For
this reason, GM detectors are often used for inexpensive survey meters.
GM detectors have high efficiencies for detecting charged particles that penetrate the walls of the detectors; almost every such particle reaching the interior of a
detector is counted. Many GM detectors are equipped with thin windows to allow
beta particles and conversion electrons to reach the gas and be detected. Very weak
charged particles, such as the beta particles emitted by tritium (3H, Ernax 18 keV),
which is extensively used in biomedical research, cannot penetrate the windows;
therefore, contamination by 3H cannot be detected with a GM survey meter. Flat,
thin-window GM detectors, called pancake-type detectors, are very useful for finding radioactive contamination (Fig. 17-7).
In general, GM survey meters are very inefficient detectors of x-rays and gamma
rays, which tend to pass through the gas without interaction. Most of those that are
detected have interacted with the walls of the detectors, with the resultant electrons
scattered into the gas inside the detectors.
The size of the voltage pulse from a GM tube is independent of the energy deposited in the detector by the interaction causing the pulse: an interaction that deposits
1 keV causes a voltage pulse of the same size as one caused by an interaction that
deposits 1 MeV. Therefore, GM detectors cannot be used as spectrometers or precise
dose-rate meters. Many portable GM survey meters display measurements in units
642
FIGURE 17-7 Portable GM survey meter with a thin-window pancake probe. In the upper left (A),
a survey for radioactive contamination is being performed. In the lower left (B), the range knob below the display is set to X10 and so the red needle on the meter indicates a count rate of about 3,500 counts per minute
(cpm). The thin window of the GM probe (C) is designed to permit beta particles and conversion electrons
whose energies exceed about 45 keV to reach the sensitive volume inside the tube, and the large surface area
of the detector reduces the time needed to survey a surface.
of milliroentgens per hour. However, the GM counter cannot truly measure exposure rates, and so its reading must be considered only an approximation. If a GM
survey meter is calibrated to indicate exposure rate for 662-keV gamma rays from
137
Cs (commonly used for calibrations), it may overrespond by as much as a factor of
5 for photons of lower energies, such as 80 keV. If an accurate measurement of exposure rate is required, an air-filled ionization chamber survey meter should be used.
This overresponse of a GM tube to low-energy x-rays and gamma rays can be
partially corrected by placing a thin layer of a material with a moderately high atomic
number (e.g., tin) around the detector. The increasing attenuation coefficient of the
material (due to the photoelectric effect) with decreasing photon energy significantly
flattens the energy response of the detector. Such GM tubes are called energy-compensated detectors. The disadvantage of an energy-compensated detector is that its
sensitivity to lower energy photons is substantially reduced and its energy threshold,
below which photons cannot be detected at all, is increased. Energy-compensated
GM detectors commonly have windows that can be opened to expose the thin tube
walls so that high-energy beta particles and low-energy photons can be detected.
GM detectors suffer from extremely long dead times, ranging from tens to hundreds of microseconds. For this reason, GM counters are seldom used when accurate
643
measurements are required of count rates greater than a few hundred counts per
second. A portable GM survey meter may become paralyzed in a very high radiation
field and yield a reading of zero. Ionization chamber instruments should always be
used to measure high intensity x-ray and gamma ray fields.
17.3 Scintillation Detectors
Basic Principles
Scintillators are materials that emit visible light or ultraviolet radiation after the interaction of ionizing radiation with the material. Scintillators are the oldest type of
radiation detectors; Roentgen discovered x-radiation and the fact that x-rays induce
scintillation in barium platinocyanide in the same fortuitous experiment. Scintillators
are used in conventional film-screen radiography, many direct digital radiographic
image receptors, fluoroscopy, scintillation cameras, CT scanners, and positron
emission tomography (PET) scanners.
Although the light emitted from a single interaction can be seen if the viewers eyes
are dark adapted, most scintillation detectors incorporate a means of signal amplification. In conventional film-screen radiography, photographic film is used to amplify
and record the signal. In other applications, electronic devices such as photomultiplier
tubes (PMTs), photodiodes, or image-intensifier tubes convert the light into electrical
signals. PMTs and image-intensifier tubes amplify the signal as well. However, most
photodiodes do not provide amplification; if amplification of the signal is required, it
must be provided by an electronic amplifier. A scintillation detector consists of a scintillator and a device, such as a PMT, that converts the light into an electrical signal.
When ionizing radiation interacts with a scintillator, electrons are raised to an
excited energy level. Ultimately, these electrons fall back to a lower energy state,
with the emission of visible light or ultraviolet radiation. Most scintillators have
more than one mode for the emission of visible light or ultraviolet radiation, and
each mode has its characteristic decay constant. Luminescence is the emission of light
after excitation. Fluorescence is the prompt emission of light, whereas phosphorescence
(also called afterglow) is the delayed emission of light. When scintillation detectors are operated in current mode, the prompt signal from an interaction cannot be
separated from the phosphorescence caused by previous interactions. When a scintillation detector is operated in pulse mode, afterglow is less important because electronic circuits can separate the rapidly rising and falling components of the prompt
signal from the slowly decaying delayed signal resulting from previous interactions.
It is useful, before discussing actual scintillation materials, to consider properties
that are desirable in a scintillator.
1. The conversion efficiency, the fraction of deposited energy that is converted into
light or ultraviolet (UV) radiation, should be high. (Conversion efficiency should
not be confused with detection efficiency.)
2. For many applications, the decay times of excited states should be short. (Light or
UV radiation is emitted promptly after an interaction.)
3. The material should be transparent to its own emissions. (Most emitted light or
UV radiation escapes reabsorption.)
4. The frequency spectrum (color) of emitted light or UV radiation should match the
spectral sensitivity of the light receptor (PMT, photodiode, or film).
5. If used for x-ray and gamma-ray detection, the attenuation coefficient (m) should
be large, so that detectors made of the scintillator have high detection efficiencies.
644
Materials with large atomic numbers and high densities have large attenuation
coefficients.
6. The material should be rugged, unaffected by moisture, and inexpensive to
manufacture.
In all scintillators, the amount of light emitted after an interaction increases with
the energy deposited by the interaction. Therefore, scintillators may be operated in
pulse mode as spectrometers. When a scintillator is used for spectroscopy, its energy
resolution (ability to distinguish between interactions depositing different energies)
is primarily determined by its conversion efficiency. A high conversion efficiency is
required for superior energy resolution.
There are several categories of materials that scintillate. Many organic compounds
exhibit scintillation. In these materials, the scintillation is a property of the molecular
structure. Solid organic scintillators are used for timing experiments in particle physics because of their extremely prompt light emission. Organic scintillators include the
liquid scintillation fluids that are used extensively in biomedical research. Samples
containing radioactive tracers such as 3H, 14C, and 32P are mixed in vials with liquid
scintillators, and the light flashes are detected and counted by PMTs and associated
electronic circuits. Organic scintillators are not used for medical imaging because the
low atomic numbers of their constituent elements and their low densities make them
poor x-ray and gamma-ray detectors. When photons in the diagnostic energy range
do interact with organic scintillators, it is primarily by Compton scattering.
There are also many inorganic crystalline materials that exhibit scintillation. In these
materials, the scintillation is a property of the crystalline structure: if the crystal is dissolved, the scintillation ceases. Many of these materials have much larger average atomic
numbers and higher densities than organic scintillators and therefore are excellent photon
detectors. They are widely used for radiation measurements and imaging in radiology.
Most inorganic scintillation crystals are deliberately grown with trace amounts of
impurity elements called activators. The atoms of these activators form preferred sites
in the crystals for the excited electrons to return to the ground state. The activators
modify the frequency (color) of the emitted light, the promptness of the light emission, and the proportion of the emitted light that escapes reabsorption in the crystal.
Inorganic Crystalline Scintillators in Radiology

No one scintillation material is best for all applications in radiology. Sodium iodide
activated with thallium [NaI(Tl)] is used for most nuclear medicine applications. It is
coupled to PMTs and operated in pulse mode in scintillation cameras, thyroid probes,
and gamma well counters. Its high content of iodine (Z 53) and high density provide a high photoelectric absorption probability for x-rays and gamma rays emitted by
common nuclear medicine radiopharmaceuticals (70 to 365 keV). It has a very high
conversion efficiency; approximately 13% of deposited energy is converted into light.
Because a light photon has an energy of about 3 eV, approximately one light photon is
emitted for every 23 eV absorbed by the crystal. This high conversion efficiency gives
it a very good energy resolution. It emits light very promptly (decay constant, 250ns),
permitting it to be used in pulse mode at interaction rates greater than 100,000/s.
Very large crystals can be manufactured; for example, the rectangular crystals of one
modern scintillation camera are 59 cm (23 inches) long, 44.5 cm (17.5 inches) wide,
and 0.95 cm thick. Unfortunately, NaI(Tl) crystals are fragile; they crack easily if
struck or subjected to rapid temperature change. Also, they are hygroscopic (i.e., they
absorb water from the atmosphere) and therefore must be hermetically sealed.
Positron emission tomography (PET), discussed in Chapter 19, requires high
detection efficiency for 511-keV annihilation photons and a prompt signal from each
645
interaction because the signals must be processed in pulse mode at high interaction
rates. PET detectors are thick crystals of high-density, high atomic number scintillators optically coupled to PMTs. For many years, bismuth germanate (Bi4Ge3O12, often
abbreviated as BGO) was the preferred scintillator. The high atomic number of
bismuth (Z 83) and the high density of the crystal yield a high intrinsic efficiency
for the 511-keV positron annihilation photons. The primary component of the light
emission is sufficiently prompt (decay constant, 300 ns) for PET. NaI(Tl) was used
in early and some less-expensive PET scanners. Today, lutetium oxyorthosilicate
(Lu2SiO4O, abbreviated LSO), lutetium yttrium oxyorthosilicate (LuxY2-xSiO4O,
abbreviated LYSO), and gadolinium oxyorthosilicate (Gd2SiO4O, abbreviated GSO),
all activated with cerium, are used in newer PET scanners. Their densities and effective atomic numbers are similar to those of BGO, but their conversion efficiencies are
much larger and they emit light much more promptly.
Calcium tungstate (CaWO4) was used for many years in intensifying screens
in film-screen radiography. It was largely replaced by rare-earth phosphors, such
as gadolinium oxysulfide activated with terbium. The intensifying screen is an
application of scintillators that does not require very prompt light emission, because
the film usually remains in contact with the screen for at least several seconds after
exposure. Cesium iodide activated with thallium is used as the phosphor layer of
many indirect-detection thin-film transistor radiographic and fluoroscopic image
receptors, described in Chapters7 and 9. Cesium iodide activated with sodium is
used as the input phosphor and zinc cadmium sulfide activated with silver is used as
the output phosphor of image-intensifier tubes in fluoroscopes.
Scintillators coupled to photodiodes are used as the detectors in CT scanners, as
described in Chapter 10. The extremely high x-ray flux experienced by the detectors
necessitates current mode operation to avoid dead-time effects. With the rotational
speed of CT scanners as high as three rotations per second, the scintillators used in
CT must have very little afterglow. Cadmium tungstate and gadolinium ceramics are
scintillators used in CT. Table 17-1 lists the properties of several inorganic crystalline
scintillators of importance in radiology and nuclear medicine.
Conversion of Light into an Electrical Signal

Photomultiplier Tubes
PMTs perform two functionsconversion of ultraviolet and visible light photons
into an electrical signal and signal amplification, on the order of millions to billions.
As shown in Figure 17-8, a PMT consists of an evacuated glass tube containing a
photocathode, typically 10 to 12 electrodes called dynodes, and an anode. The photocathode is a very thin electrode, located just inside the glass entrance window
of the PMT, which emits electrons when struck by visible light. Photocathodes are
inefficient; approximately one electron is emitted from the photocathode for every
five UV or light photons incident upon it. A high-voltage power supply provides a
voltage of approximately 1,000 V, and a series of resistors divides the voltage into
equal increments. The first dynode is given a voltage of about 100 V with respect to
the photocathode; successive dynodes have voltages that increase by approximately
100 V per dynode. The electrons emitted by the photocathode are attracted to the
first dynode and are accelerated to kinetic energies equal to the potential difference
between the photocathode and the first dynode. (If the potential difference is 100 V,
the kinetic energy of each electron is 100 eV.) When these electrons strike the first
dynode, about five electrons are ejected from the dynode for each electron hitting it.
These electrons are then attracted to the second dynode, reaching kinetic energies
equal to the potential difference between the first and second dynodes, and causing
7.90
6.12
83, 32, 8
71, 14, 8
55, 53
55, 53
30, 48, 16
48, 74, 8
20, 74, 8
64, 8, 16
NaI(Tl)
Bi4Ge3O12
Lu2SiO4O(Ce)
CsI(Na)
CsI(Tl)
ZnCdS(Ag)
CdWO4
CaWO4
Gd2O2S(Tb)
475
550
420
420
480
415
WAVELENGTH
OF MAXIMAL
EMISSION (nm)
1418
40
45b
85
75
1214
100
CONVERSION
EFFICIENCYa (%)
560
0.920
14
0.1 @ 3 ms
Radiographic screens
Radiographic screens
Computed tomographic (CT)

scanners
Output phosphor of imageintensifier tubes
Thin-film transistor radiographic

and fluoroscopic image receptors
1.0
PET scanners
PET scanners
Scintillation cameras
USES
Input phosphor of image-intensifier

tubes
0.55 @ 6 ms
0.005 @ 3 ms
0.35 @ 6 ms
AFTERGLOW (%)
0.63
40
0.3
0.25
DECAY CONSTANT
(s)
Relative to NaI(Tl), using a PMT to measure light.

The light emitted by CsI(Tl) does not match the spectral sensitivity of PMTs very well; its conversion efficiency is much larger if measured with a photodiode.
Source: Data on NaI(Tl), BGO, CsI(Na), CsI(Tl), and CdWO4 courtesy of Saint-Gobain Crystals, Hiram, OH. Data on LSO from Ficke DC, Hood JT, Ter-Pogossian MM. A spheroid positron
emission tomograph for brain imaging: a feasibility study. JNM 1996: 37:1222.
7.34
4.51
4.51
7.4
7.13
3.67
11, 53
MATERIAL
DENSITY
(g/cm3)
ATOMIC
NUMBERS
TABLE 17-1 INORGANIC SCINTILLATORS USED IN MEDICAL IMAGING
646

Focusing
Electrode
100
volts
Electron
Multiplication
300
volts
Light or UV Photon
Semitransparent
Photocathode
200
volts
Photoelectron
Dynodes
400
volts
647
Evacuated
Glass
Envelope
Electrical
Connectors
500
volts
600
volts
Anode
FIGURE 17-8 Diagram of a PMT showing the main components (photocathode, focusing electrode, dynodes,
and anode) and illustrating the process of electron multiplication. Actual PMTs typically have 10 to 12dynodes.
about five electrons to be ejected from the second dynode for each electron hitting
it. This process continues down the chain of dynodes, with the number of electrons
being multiplied by a factor of 5 at each stage. The total amplification of the PMT is
the product of the individual multiplications at each dynode. If a PMT has ten dynodes and the amplification at each stage is 5, the total amplification will be
5 5 5 5 5 5 5 5 5 5 510 10,000,000
The amplification can be adjusted by changing the voltage applied to the PMT.
When a scintillator is coupled to a PMT, an optical coupling material is placed
between the two components to minimize reflection losses. The scintillator is usually
surrounded on all other sides by a highly reflective material, often magnesium oxide
powder.
Photodiodes
Photodiodes are semiconductor diodes that convert light into electrical signals. (The
principles of operation of semiconductor diodes are discussed later in this chapter.) In
use, photodiodes are reverse biased. Reverse bias means that the voltage is applied with
the polarity such that essentially no electrical current flows. When the photodiode is
exposed to light, an electrical current is generated that is proportional to the intensity of the light. Photodiodes are sometimes used with scintillators instead of PMTs.
Photodiodes produce more electrical noise than PMTs do, but they are smaller and
less expensive. Most photodiodes, unlike PMTs, do not amplify the signal. However, a
type of photodiode called an avalanche photodiode does provide signal amplification,
although not as much as a PMT. Photodiodes coupled to CdWO4 or other scintillators
are used in current mode in CT scanners. Photodiodes are also essential components
of indirect-detection thin-film transistor radiographic and fluoroscopic image receptors, which use scintillators to convert x-ray energy into light.
Scintillators with Trapping of Excited Electrons

In most applications of scintillators, the prompt emission of light after an interaction
is desirable. However, there are inorganic scintillators in which electrons become
trapped in excited states after interactions with ionizing radiation. These trapped
electrons can be released by heating or exposure to light; the electrons then fall to
their ground state with the emission of light, which can be detected by a PMT or
648
other sensor. These trapped electrons, in effect, store information about the radiation
exposure. Such scintillators can be used for dosimetry or for radiographic imaging.
Thermoluminescent Dosimeters and Optically

StimulatedLuminescent Dosimeters
As mentioned above, scintillators with electron trapping can be used for dosimetry.
In the case of thermoluminescent dosimeters (TLDs), to read the signal after exposure
to ionizing radiation, a sample of TLD material is heated, the light is detected and
converted into an electrical signal by a PMT, and the resultant signal is integrated
and displayed. The amount of light emitted by the TLD increases with the amount
of energy absorbed by the TLD, but may deviate from proportionality, particularly at
higher doses. After the TLD has been read, it may be baked in an oven to release the
remaining trapped electrons and reused.
Lithium fluoride (LiF) is one of the most useful TLD materials. It is commercially
available in forms with different trace impurities (Mg and Ti or Mg, Cu, and P), giving differences in properties such as sensitivity and linearity of response with dose.
LiF has trapping centers that exhibit almost negligible release of trapped electrons at
room temperature, so there is little loss of information with time from exposure to
the reading of the TLD. The effective atomic number of LiF is close to that of tissue,
so the amount of light emission is almost proportional to the tissue dose over a wide
range of x-ray and gamma-ray energies. It is commonly used instead of photographic
film for personnel dosimetry.
In optically stimulated luminescense (OSL), the trapped excited electrons are
released by exposure to light, commonly produced by a laser, of a frequency optimal for
releasing the trapped electrons. The most commonly used OSL material is aluminum
oxide (Al2O3) activated with a small amount of carbon. The effective atomic number of
aluminum oxide is significantly higher than that of soft tissue, and so dose to this material is not proportional to dose to soft tissue over the full range of energies used in medical imaging. Methods for compensating for this effect are discussed in Chapter 21.
Photostimulable Phosphors
Photostimulable phosphors (PSPs), like TLDs, are scintillators in which a fraction of the
excited electrons become trapped. PSP plates are used in radiography as image receptors, instead of film-screen cassettes. Although the trapped electrons could be released
by heating, a laser is used to scan the plate and release them. The electrons then fall to the
ground state, with the emission of light. Barium fluorohalide activated with europium
is commonly used for PSP imaging plates. In this material, the wavelength that is most
efficient in stimulating luminescence is in the red portion of the spectrum, whereas the
stimulated luminescence itself is in the blue-violet portion of the spectrum. The stimulated emissions are converted into an electrical signal by PMTs. After the plate is read by
the laser, it may be exposed to light to release the remaining trapped electrons that can
be reused. The use of PSPs in radiography has been discussed further in Chapter 7.
17.4 Semiconductor Detectors

Semiconductors are crystalline materials whose electrical conductivities are less than
those of metals but more than those of crystalline insulators. Silicon and germanium
are common semiconductor materials.
In crystalline materials, electrons exist in energy bands, separated by forbidden
gaps. In metals (e.g., copper), the least tightly bound electrons exist in a partially
occupied band, called the conduction band. The conduction-band electrons are
mobile, providing high electrical conductivity. In an insulator or a semiconductor,
649
Conduction band
(empty)
Electron
energy
forbidden energy gap

(~ 10 eV)
Conduction band
forbidden energy gap (~ 1 eV)
Valence band
(full)
Valence band
Insulator
Semiconductor
FIGURE 17-9 Energy band structure of a crystalline insulator and a semiconductor material.
the valence electrons exist in a filled valence band. In semiconductors, these valenceband electrons participate in covalent bonds and so are immobile. The next higher
energy band, the conduction band, is empty of electrons. However, if an electron
is placed in the conduction band, it is mobile, as are the upper band electrons in
metals. The difference between insulators and semiconductors is the magnitude of the
energy gap between the valence and conduction bands. In insulators, the band gap
is greater than 5 eV, whereas in semiconductors, it is about 1 eV or less (Fig.17-9).
In semiconductors, valence-band electrons can be raised to the conduction band by
ionizing radiation, visible light or ultraviolet radiation, or thermal energy.
When a valence-band electron is raised to the conduction band, it leaves behind a
vacancy in the valence band. This vacancy is called a hole. Because a hole is the absence
of an electron, it is considered to have a net positive charge, equal but opposite to that
of an electron. When another valence-band electron fills the hole, a hole is created
at that electrons former location. Thus, holes behave as mobile positive charges in
the valence band even though positively charged particles do not physically move in
the material. The hole-electron pairs formed in a semiconductor material by ionizing
radiation are analogous to the ion pairs formed in a gas by ionizing radiation.
A crystal of a semiconductor material can be used as a radiation detector. A voltage is placed between two terminals on opposite sides of the crystal. When ionizing
radiation interacts with the detector, electrons in the crystal are raised to an excited
state, permitting an electrical current to flow, similar to a gas-filled ionization chamber. Unfortunately, the radiation-induced current, unless it is very large, is masked by
a larger current induced by the applied voltage.
To reduce the magnitude of the voltage-induced current so that the signal from
radiation interactions can be detected, the semiconductor crystal is doped with a
trace amount of impurities so that it acts as a diode (see earlier discussion of types of
detectors). The impurity atoms fill sites in the crystal lattice that would otherwise be
occupied by atoms of the semiconductor material. If atoms of the impurity material
have more valence electrons than those of the semiconductor material, the impurity
atoms provide mobile electrons in the conduction band. A semiconductor material containing an electron-donor impurity is called an n-type material (Fig. 17-10).
N-type material has mobile electrons in the conduction band. On the other hand, an
impurity with fewer valence electrons than the semiconductor material provides sites
in the valence band that can accept electrons. When a valence-band electron fills one
of these sites, it creates a hole at its former location. Semiconductor material doped
with a hole-forming impurity is called p-type material. P-type material has mobile
holes in the valence band.
650
Si
Si
Si
Si
Si
Si
Si
Si
Si
Si
Si
Si
Si
p-type material
Si
Si
Si
n-type material
FIGURE 17-10 P-type and n-type impurities in a crystal of a semiconductor material, silicon in this example.
N-type impurities provide mobile electrons in the conduction band, whereas p-type impurities provide acceptor
sites in the valence band. When filled by electrons, these acceptor sites create holes that act as mobile positive
charges. (Si, silicon; B, boron; P, phosphorus.)
A semiconductor diode consists of a crystal of semiconductor material with a region

of n-type material that forms a junction with a region of p-type material (Fig.17-11).
If an external voltage is applied with the positive polarity on the p-type side of the
diode and the negative polarity on the n-type side, the holes in the p-type material
and the mobile conduction-band electrons of the n-type material are drawn to the
junction. There, the mobile electrons fall into the valence band to fill holes. Applying
an external voltage in this manner is referred to as forward bias. Forward bias permits
a current to flow with little resistance.
On the other hand, if an external voltage is applied with the opposite polaritythat
is, with the negative polarity on the p-type side of the diode and the positive polarity
on the n-type sidethe holes in the p-type material and the mobile conductionband electrons of the n-type material are drawn away from the junction. Applying
the external voltage in this polarity is referred to as reverse bias. Reverse bias draws
the charge carriers away from the n-p junction, forming a region depleted of current
carriers. Very little electrical current flows when a diode is reverse biased.
FIGURE 17-11 Semiconductor diode. When
no bias is applied, a few holes migrate into
the n-type material and a few conductionband electrons migrate into the p-type material. With forward bias, the external voltage
is applied with the positive polarity on the
p-side of the junction and negative polarity
on the n-side, causing the charge carriers to
be swept into the junction and a large current
to flow. With negative bias, the charge carriers
are drawn away from the junction, creating a
region depleted of charge carriers that acts as
a solid-state ion chamber.
n-type
No external
voltage applied
Forward bias
(current flows)
Reverse
bias
p-type
+
+ +

power supply
depletion
region
power supply
+
+
+
+
+ +
+ +
+ +
651
A reverse-biased semiconductor diode can be used to detect visible light and

UV radiation or ionizing radiation. The photons of light or ionization and excitation
produced by ionizing radiation can excite lower energy electrons in the depletion
region of the diode to higher energy bands, producing hole-electron pairs. The electrical field in the depletion region sweeps the holes toward the p-type side and the
conduction-band electrons toward the n-type side, causing a momentary pulse of
current to flow after the interaction.
Photodiodes are semiconductor diodes that convert light into an electrical current.
As mentioned previously, they are used in conjunction with scintillators as detectors in
CT scanners. Scintillation-based thin-film transistor radiographic and fluoroscopic image
receptors, discussed in Chapter 7, incorporate a photodiode in each detector element.
Semiconductor detectors are semiconductor diodes designed for the detection of
ionizing radiation. The amount of charge generated by an interaction is proportional
to the energy deposited in the detector by the interaction; therefore, semiconductor
detectors are spectrometers. Because thermal energy can also raise electrons to the
conduction band, many types of semiconductor detectors used for x-ray and gammaray spectroscopy must be cooled with liquid nitrogen.
The energy resolution of germanium semiconductor detectors is greatly superior
to that of NaI(Tl) scintillation detectors. Liquid nitrogencooled germanium detectors are widely used for the identification of individual gamma rayemitting radionuclides in mixed radionuclide samples because of their superb energy resolution.
Semiconductor detectors are seldom used for medical imaging devices because of
high expense, because of low quantum detection efficiencies in comparison to scintillators such as NaI(Tl) (Z of iodine 53, Z of germanium 32, Z of silicon14),
because they can be manufactured only in limited sizes, and because many such
devices require cooling.
Efforts are being made to develop semiconductor detectors of higher atomic
number than germanium that can be operated at room temperature. A leading candidate to date is cadmium zinc telluride (CZT). A small-field-of-view nuclear medicine
camera using CZT detectors has been developed.
A layer of a semiconductor material, amorphous selenium (Z 34), is used in
some radiographic image receptors, including those in some mammography systems. The selenium is commonly referred to as a photoconductor. In these image
receptors, the selenium layer is electrically coupled to a rectangular array of thin film
transistor detector elements which collect and store the mobile electrical charges produced in the selenium by x-ray interactions during image acquisition. These image
receptors are discussed in Chapters 7 and 8.
17.5 Pulse Height Spectroscopy

Many radiation detectors, such as scintillation detectors, semiconductor detectors, and
proportional counters, produce electrical pulses whose amplitudes are proportional
to the energies deposited in the detectors by individual interactions. Pulse height analyzers (PHAs) are electronic systems that may be used with these detectors to perform
pulse height spectroscopy and energy-selective counting. In energy-selective counting, only interactions that deposit energies within a certain energy range are counted.
Energy-selective counting can be used to reduce the effects of background radiation,
to reduce the effects of scatter, or to separate events caused by different radionuclides
in a sample containing multiple radionuclides. Two types of PHAs are single-channel
analyzers (SCAs) and multichannel analyzers (MCAs). MCAs determine spectra much
652
more efficiently than do SCA systems, but they are more expensive. Pulse height
discrimination circuits are incorporated in scintillation cameras and other nuclear
medicine imaging devices to reduce the effects of scatter on the images.
Single-Channel Analyzer Systems

Function of a Single-Channel Analyzer System
Figure 17-12 depicts an SCA system. Although the system is shown with an NaI(Tl)
crystal and PMT, it could be used with any pulse-mode spectrometer. The high-voltage
power supply typically provides 800 to 1,200 V to the PMT. The series of resistors
divides the total voltage into increments that are applied to the dynodes and anode of the
PMT. Raising the voltage increases the magnitude of the voltage pulses from the PMT.
The detector is often located some distance from the majority of the electronic
components. The pulses from the PMT are usually routed to a preamplifier (preamp), which is connected to the PMT by as short a cable as possible. The function
of the preamp is to amplify the voltage pulses further, so as to minimize distortion
and attenuation of the signal during transmission to the remainder of the system.
The pulses from the preamp are routed to the amplifier, which further amplifies the
pulses and modifies their shapes. The gains of most amplifiers are adjustable.
The pulses from the amplifier then proceed to the SCA. The user is allowed to set
two voltage levels, a lower level and an upper level. If a voltage pulse whose amplitude is less than the lower level or greater than the upper level is received from the
amplifier, the SCA does nothing. If a voltage pulse whose amplitude is greater than
the lower level but less than the upper level is received from the amplifier, the SCA
produces a single logic pulse. A logic pulse is a voltage pulse of fixed amplitude and
duration. Figure 17-13 illustrates the operation of an SCA. The counter counts the
logic pulses from the SCA for a time interval set by the timer.
Many SCAs permit the user to select the mode by which the two knobs set the
lower and upper levels. In one mode, usually called LL/UL mode, one knob directly
sets the lower level and the other sets the upper level. In another mode, called window
mode, one knob (often labeled E or energy) sets the midpoint of the range of acceptable pulse heights and the other knob (often labeled E or window) sets the range of
voltages around this value. In this mode, the lower level voltage is E E/2 and the
upper level voltage is E E/2. (In some SCAs, the range of acceptable pulse heights
is from E to E E.) Window mode is convenient for plotting a spectrum.
FIGURE 17-12 SCA system
with NaI(Tl) detector and PMT.
magnesium oxide
reflector
Nal (TI)
crystal
preamplifier
PMT
aluminum
can
high voltage
power supply +
amplifier
SCA
E
counter
timer
FIGURE 17-13 Illustration of the function

of an SCA. Energy discrimination occurs by
rejection of pulses above or below the energy
window set by the operator.
Input to SCA
Voltage
653
E
Time
Voltage
Output from SCA
Time
Plotting a Spectrum Using a Single-Channel Analyzer

To obtain the pulse height spectrum of a sample of radioactive material using an SCA
system, the SCA is placed in window mode, the E setting is set to zero, and a small
window (E) setting is selected. A series of counts is taken for a fixed length of time per
count, with the E setting increased before each count but without changing the window
setting. Each count is plotted on graph paper as a function of baseline (E) setting.
Energy Calibration of a Single-Channel Analyzer System

On most SCAs, each of the two knobs permits values from 0 to 1,000 to be selected.
By adjusting the amplification of the pulses reaching the SCAeither by changing
the voltage produced by the high-voltage power supply or by changing the amplifier
gainthe system can be calibrated so that these knob settings directly indicate keV.
A 137Cs source, which emits 662-keV gamma rays, is usually used. A narrow window is set, centered about a setting of 662. For example, the SCA may be placed into
LL/UL mode with lower level value of 655 and an upper level value of 669 selected.
Then the voltage produced by the high-voltage power supply is increased in steps,
with a count taken after each step. The counts first increase and then decrease. When
the voltage that produces the largest count is selected, the two knobs on the SCA
directly indicate keV. This procedure is called peaking the SCA system.
Multichannel Analyzer Systems

An MCA system permits an energy spectrum to be automatically acquired much more
quickly and easily than does an SCA system. Figure 17-14 is a diagram of a counting system using an MCA. The detector, high-voltage power supply, preamp, and
amplifier are the same as were those described for SCA systems. The MCA consists
of an analog-to-digital converter (ADC), a memory containing many storage locations called channels, control circuitry, and a timer. The memory of an MCA typically
ranges from 256 to 8,192 channels, each of which can store a single integer. When
the acquisition of a spectrum begins, all of the channels are set to zero. When each
voltage pulse from the amplifier is received, it is converted into a binary digital signal, the value of which is proportional to the amplitude of the analog voltage pulse.
(ADCs are discussed in Chapter 5.) This digital signal designates a particular channel
in the MCAs memory. The number stored in that channel is then incremented by 1.
654
FIGURE 17-14 Diagram of an MCA

system with NaI(Tl) detector.
Nal (TI)
crystal
preamplifier
PMT
high voltage
power supply +
Multi-channel
Analyzer
amplifier
As many pulses are processed, a spectrum is generated in the memory of the MCA.
Figure 17-15 illustrates the operation of an MCA. Today, most MCAs are interfaced
to digital computers that store, process, and display the resultant spectra.
x-ray and Gamma-Ray Spectroscopy with

SodiumIodideDetectors
analog
voltage
pulses
ADC
digital
signal
Memory
0 0
0 1
1st pulse
ADC
ADC
2nd pulse
3rd pulse
ADC
After 12 pulses have

been analyzed:
0 0
2 3 4 5 6 7
channel number
0 0
1 0
0 1
6 7
0 0
1 0
0 1
6 7
0 0
2 0
0 1
6 7
1 1
4 2
0 1
6 7
no. of counts
x-ray and gamma-ray spectroscopy is best performed with semiconductor detectors because of their superior energy resolution. However, high detection efficiency
Display
energy
FIGURE 17-15 Acquisition of a spectrum by an MCA. The digital signal produced by the ADC is a binary
signal, as described in Chapter 5. After the analog pulses are digitized by the ADC, they are sorted into bins
(channels) by height, forming an energy spectrum. Although this figure depicts an MCA with 8 channels,
actual MCAs have as many as 8,192 channels.
655
is more important than ultrahigh energy resolution for most nuclear medicine
applications, so most spectroscopy systems in nuclear medicine use NaI(Tl) crystals
coupled to PMTs.
Interactions of Photons with a Spectrometer

There are a number of mechanisms by which an x-ray or gamma ray can deposit
energy in the detector, several of which deposit only a fraction of the incident photon
energy. As illustrated in Figure 17-16, an incident photon can deposit its full energy
by a photoelectric interaction (A) or by one or more Compton scatters followed by
a photoelectric interaction (B). However, a photon will deposit only a fraction of its
energy if it interacts by Compton scattering and the scattered photon escapes the
detector (C). In that case, the energy deposited depends on the scattering angle, with
larger angle scatters depositing larger energies. Even if the incident photon interacts
by the photoelectric effect, less than its total energy will be deposited if the innershell electron vacancy created by the interaction results in the emission of a characteristic x-ray that escapes the detector (D).
Most detectors are shielded to reduce the effects of natural background radiation
and nearby radiation sources. Figure 17-16 shows two ways by which an x-ray or
gamma-ray interaction in the shield of the detector can deposit energy in the detector. The photon may Compton scatter in the shield, with the scattered photon striking the detector (E), or a characteristic x-ray from the shield may interact with the
detector (F).
Most interactions of x-rays and gamma rays with an NaI(Tl) detector are with
iodine atoms, because iodine has a much larger atomic number than sodium does.
Although thallium has an even larger atomic number, it is only a trace impurity.
Spectrum of Cesium-137
The spectrum of 137Cs is often used to introduce pulse height spectroscopy because of
the simple decay scheme of this radionuclide. As shown at the top of Figure 17-17,
137
Cs decays by beta particle emission to 137mBa, whose nucleus is in an excited state. The
137m
Ba nucleus attains its ground state by the emission of a 662-keV gamma ray 90% of
the time. In 10% of the decays, a conversion electron is emitted instead of a gamma ray.
The conversion electron is usually followed by the emission of an approximately 32-keV
K-shell characteristic x-ray as an outer-shell electron fills the inner-shell vacancy.
FIGURE 17-16 Interactions of x-rays
and gamma rays with an NaI(Tl) detector.
See text for description.
Lead Shield
A
Source
Nal(TI)
PMT
C
D
Lead Shield
656

Cesium-137
Beta-minus decay
137Cs
55
1
137m
661.6 keV
Ba
0.0
100
137
Ba
56
662 keV
32 keV
32 keV
E
Number of Counts
Relative Number of Photons
Stable
662 keV
F
D
C
B
100 200 300 400 500 600 700

Energy (keV)
100
200
300 400 500

Energy (keV)
600
700
FIGURE 17-17 Decay scheme of 137Cs (top), actual energy spectrum (left), and pulse height spectrum
obtained using an NaI(Tl) scintillation detector (right). See text for description of pulse height spectrum. (A)
photopeak, due to complete absorption of 662-keV gamma rays in the crystal; (B) Compton continuum; (C)
Compton edge; (D) backscatter peak; (E) barium x-ray photopeak; and (F) photopeak caused by absorption of
lead K-shell x-rays (72 to 88 keV) from the shield.
In the left in Figure 17-17 is the actual energy spectrum of 137Cs, and on the
right is its pulse height spectrum obtained with the use of an NaI(Tl) detector. There
are two reasons for the differences between the spectra. First, there are a number of
mechanisms by which an x-ray or gamma ray can deposit energy in the detector, several of which deposit only a fraction of the incident photon energy. Second, there are
random variations in the processes by which the energy deposited in the detector is
converted into an electrical signal. In the case of an NaI(Tl) crystal coupled to a PMT,
there are random variations in the fraction of deposited energy converted into light,
the fraction of the light that reaches the photocathode of the PMT, and the number
of electrons ejected from the back of the photocathode per unit energy deposited by
the light. These factors cause random variations in the size of the voltage pulses produced by the detector, even when the incident x-rays or gamma rays deposit exactly
the same energy. The energy resolution of a spectrometer is a measure of the effect of
these random variations on the resultant spectrum.
In the pulse height spectrum of 137Cs, on the right in Figure 17-17, the photopeak (A) is caused by interactions in which the energy of an incident 662-keV
photon is entirely absorbed in the crystal. This may occur by a single photoelectric
interaction or by one or more Compton scattering interactions followed by a photoelectric interaction. The Compton continuum (B) is caused by 662-keV photons
that scatter in the crystal, with the scattered photons escaping the crystal. Each portion of the continuum corresponds to a particular scattering angle. The Compton
edge (C) is the upper limit of the Compton continuum. The backscatter peak (D)
is caused by 662-keV photons that scatter from the shielding around the detector
into the detector. The barium x-ray photopeak (E) is a second photopeak caused by
657
the absorption of barium K-shell x-rays (31 to 37 keV), which are emitted after the
emission of conversion electrons. Another photopeak (F) is caused by lead K-shell
x-rays (72 to 88 keV) from the shield.
Spectrum of Technetium-99m
The decay scheme of 99mTc is shown at the top in Figure 17-18. 99mTc is an isomer
of 99Tc that decays by isomeric transition to its ground state, with the emission of a
140.5-keV gamma ray. In 11% of the transitions, a conversion electron is emitted
instead of a gamma ray.
The pulse height spectrum of 99mTc is shown at the bottom of Figure 17-18. The
photopeak (A) is caused by the total absorption of the 140-keV gamma rays. The escape
peak (B) is caused by 140-keV gamma rays that interact with the crystal by the photoelectric effect but with the resultant iodine K-shell x-rays (28 to 33 keV) escaping the
crystal. There is also a photopeak (C) caused by the absorption of lead K-shell x-rays
from the shield. The Compton continuum is quite small, unlike the continuum in the
spectrum of 137Cs, because the photoelectric effect predominates in iodine at 140 keV.
Spectrum of Iodine-125
I decays by electron capture followed by the emission of a 35.5-keV gamma ray
(6.7% of decays) or a conversion electron. The electron capture usually leaves the
125
99mTc
43
Technetium 99m
Isomeric Transition Decay
142.7 keV
1
3
140.5
0.0
2.12 105Y
99 Tc
43
Number of Counts
C
B
20 40 60 80 100 120 140 160
Energy (keV)
FIGURE 17-18 Decay scheme of 99mTc (top) and its pulse height spectrum on an NaI(Tl) scintillation detector
(bottom). See text for details.
658
daughter nucleus with a vacancy in the K-shell. The emission of a conversion electron
usually also results in a K-shell vacancy. Each transformation of an 125I atom therefore
results in the emission, on the average, of 1.47 x-rays or gamma rays with energies
between 27 and 36 keV.
Figure 17-19 shows two pulse height spectra from 125I. The spectrum on the left
was acquired with the source located 7.5 cm from an NaI(Tl) detector, and the one on
the right was collected with the source in an NaI(Tl) well detector. The spectrum on
the left shows a large photopeak at about 30 keV, whereas the spectrum on the right
shows a peak at about 30 keV and a smaller peak at about 60 keV. The 60-keV peak
in the spectrum from the well detector is a sum peak caused by two photons simultaneously striking the detector. The sum peak is not apparent in the spectrum with the
source 7.5 cm from the detector because the much lower detection efficiency renders
unlikely the simultaneous interaction of two photons with the detector.
Performance Characteristics
Energy Resolution
The energy resolution of a spectrometer is a measure of its ability to differentiate
between particles or photons of different energies. It can be determined by irradiating
the detector with monoenergetic particles or photons and measuring the width of the
resultant peak in the pulse height spectrum. Statistical effects in the detection process
cause the amplitudes of the pulses from the detector to randomly vary about the mean
pulse height, giving the peak a gaussian shape. (These statistical effects are one reason
why the pulse height spectrum produced by a spectrometer is not identical to the actual
energy spectrum of the radiation.) A wider peak implies a poorer energy resolution.
Iodine-125
35.5 keV
Stable
125 I
53
EC1
0.0
125
52 Te
Source
7.5 cm
Nal
well detector
Nal
2 x 2 detector
30 keV
60 keV
FIGURE 17-19 Decay scheme and spectrum of

in NaI(Tl) well counter (right).
30 keV
60 keV
I source located 7.5 cm from solid NaI(Tl) crystal (left) and
125
659
max
Counts per Channel
Energy resolution = 682 635 100% = 7.1%

662
Full width at 1/2 max

(FWHM)
100
200
300
400
500
600
1/2 max
662 700
Photon Energy (keV)

FIGURE 17-20 Energy resolution of a pulse height spectrometer. The spectrum shown is that of
obtained by an NaI(Tl) scintillator coupled to a PMT.
Cs,
137
The width is usually measured at half the maximal height of the peak, as illustrated in
Figure 17-20. This is called the full width at half maximum (FWHM). The FWHM is
then divided by the pulse amplitude corresponding to the maximum of the peak:

Energy resolution
FWHM
100%
Pulse amplitude at center of peak
[17-4]
For example, the energy resolution of a 5-cm-diameter and 5-cm-thick cylindrical

NaI(Tl) crystal, coupled to a PMT and exposed to the 662-keV gamma rays of 137Cs,
is typically about 7% to 8%.
Count-Rate Effects in Spectrometers

In pulse height spectroscopy, count-rate effects are best understood as pulse pileup.
Figure 17-21 depicts the signal from a detector in which two interactions occur, separated by a very short time interval. The detector produces a single pulse, which is the
sum of the individual signals from the two interactions, having a higher amplitude
than the signal from either individual interaction. Because of this effect, operating a
pulse height spectrometer at a high count rate causes loss of counts and misplacement of counts in the spectrum.
Voltage
FIGURE 17-21 Pulse pileup. The

dashed lines represent the signals produced by two individual interactions in
the detector that occur at almost the
same time. The solid line depicts the
actual pulse produced by the detector.
This single pulse is the sum of the signals
from the two interactions.
Time
660
FIGURE 17-22 Thyroid probe system (A). (Photograph courtesy Capintec, Inc., Ramsey, New Jersey.) The
personal computer has added circuitry and software so that it functions as an MCA. An NaI(Tl) well detector
(B), discussed later in this chapter, for counting samples for radioactivity is part of this system. For thyroid
uptake tests, the radioiodine capsules are placed in a Lucite neck phantom before patient administration and
counted individually (C). At 4 to 6 h and again at about 24 h after administration, the radioactivity in the
patients neck is counted at the same distance from the probe as was the neck phantom (D).
17.6 Nonimaging Detector Applications
Sodium Iodide Thyroid Probe and Well Counter

Thyroid Probe
A nuclear medicine department typically has a thyroid probe for measuring the
uptake of 123I or 131I by the thyroid glands of patients and for monitoring the activities of 131I in the thyroid glands of staff members who handle large activities of 131I.
A thyroid probe, as shown in Figures 17-22 and 17-23, usually consists of a 5.1-cm
(2-inch)-diameter and 5.1-cm-thick cylindrical NaI(Tl) crystal coupled to a PMT,
which in turn is connected to a preamplifier. The probe is shielded on the sides and
back with lead and is equipped with a collimator so that it detects photons only from
a limited portion of the patient. The thyroid probe is connected to a high-voltage
power supply and either an SCA or an MCA system.
FIGURE 17-23 Diagram of
a thyroid probe.
Lead Shielding
Pre
Amp
PMT
Nal(TI)
Crystal
Cable
HV and pre-amp power in
pulses out
Lead Collimator
661
Thyroid Uptake Measurements

Thyroid uptake measurements may be performed using one or two capsules of 123I
or 131I sodium iodide. A neck phantom, consisting of a Lucite cylinder of diameter
similar to the neck and containing a hole parallel to its axis for a radioiodine capsule,
is required. In the two-capsule method, the capsules should have almost identical
activities. Each capsule is placed in the neck phantom and counted separately. Then,
one capsule is swallowed by the patient. The other capsule is called the standard.
Next, the emissions from the patients neck are counted, typically at 4 to 6 h after
administration, and again at 24 h after administration. Each time that the patients
thyroid is counted, the patients distal thigh is also counted for the same length of
time, to approximate nonthyroidal activity in the neck, and a background count is
obtained. All counts are performed with the NaI crystal at the same distance, typically 20 to 30 cm, from the thyroid phantom or the patients neck or thigh. (This
distance reduces the effects of small differences in distance between the detector
and the objects being counted.) Furthermore, each time that the patients thyroid is
counted, the remaining capsule is placed in the neck phantom and counted. Finally,
the uptake is calculated for each neck measurement:
Uptake
(Thyroid count Thigh count)

(Count of standard in phantom Background count)
Initial count of standard in phantom

Initial count of patient capsule in phantom
Some nuclear medicine laboratories instead use a method that requires only one
capsule. In this method, a single capsule is obtained, counted in the neck phantom,
and swallowed by the patient. As in the previous method, the patients neck and
distal thigh are counted, typically at 4 to 6 h and again at 24 h after administration.
The times of the capsule administration and the neck counts are recorded. Finally,
the uptake is calculated for each neck measurement:
(Thyroid count Thigh count)
e 0.693t/T1/2
(Count of capsule in phantom Background count)
where T1/2 is the physical half-life of the radionuclide and t is the time elapsed between
the count of the capsule in the phantom and the thyroid count. The single-capsule
method avoids the cost of the second capsule and requires fewer measurements, but
it is more susceptible to instability of the equipment, technologist error, and deadtime effects.
Sodium Iodide Well Counter

A nuclear medicine department also usually has an NaI(Tl) well counter, shown
in Figures 17-22B and 17-24. The NaI(Tl) well counter may be used for clinical
tests such as Schilling tests (a test of vitamin B12 absorption), plasma or red blood
cell volume determinations, and radioimmunoassays, although radioimmunoassays have been largely replaced by immunoassays that do not use radioactivity. The
well counter is also commonly used to assay wipe test samples to detect radioactive
contamination. The well counter usually consists of a cylindrical NaI(Tl) crystal,
either 5.1 cm (2 inches) in diameter and 5.1 cm thick or 7.6 cm (3 inches) in
diameter and 7.6 cm thick, with a hole in the crystal for the insertion of samples.
This configuration gives the counter an extremely high efficiency, permitting it
to assay samples containing activities of less than 1 nCi (103 mCi). The crystal is
662
FIGURE 17-24 Diagram of an NaI(Tl)

well counter.
Hinged
Lead-Shielded
Lid
Nal(TI)
Crystal
Lead
Shielding
PMT
Cable
HV and pre-amp power in
pulses out
Pre-Amp
coupled to a PMT, which in turn is connected to a preamplifier. A well counter

in a nuclear medicine department should have a thick lead shield, because it is
used to count samples containing nanocurie activities in the vicinity of millicurie
activities of high-energy gamma-ray emitters such as 67Ga, 111In, and 131I. The well
counter is connected to a high-voltage power supply and either an SCA or an MCA
system. Departments that perform large numbers of radioimmunoassays often use
automatic well counters, such as the one shown in Figure 17-25, to count large
numbers of samples.
Removable
Plastic Liner
Lead
Shielding
Well
Detector
Well
Detector
Lead
Shielding
Nal Well
Crystal
Photomultiplier Tube
ISOTOPE LIBRARY
1
2
3
4
5
6
7
ISOTOPE
FULL
Tc99M
I131
Co57
Co58
Cs137
I125
AVG EFF%
100 . 0(D)
75 . 1(S)
100 . 0(D)
84 . 8(S)
100 . 0(D)
9 . 7(S)
85 . 8(S)
71 EDIT ISO 72 DELETE

# NEXT PAGE 0 EXIT
SELECT FUNCTION:
WIPE SET: AREA1

Count Time: 1.0 MIN
Sample report
MDA (Minimum Detectable Activity) = (2.71 / Ts) + (3.29) * sqrt (Rb / Tb + Rb / Ts)
Location:
ISOTOPE
FULL
Tc99m
I-131
Co-58
I-125
Cr-51
In-111
Xe-133
BENCH1
BKGND
222
57
76
69
7
55
28
28
MDA
TRIGGER
20.7
17.4
16.5
134.3
10.3
25.6
13.4
13.4
2000
2000
2000
1000
1500
2000
1500
2000
NET ACTIVITY
5549
5105
0
0
0
258
84
0
UNITS
RESULT
DPM
DPM
DPM
DPM
DPM
DPM
CPM
CPM
FAIL
FAIL
PASS
PASS
PASS
PASS
PASS
PASS
FIGURE 17-25 Automatic gamma well counter. (Photograph courtesy Laboratory

Technologies, Inc.)
663
Sample Volume and Dead-Time Effects

in Sodium Iodide WellCounters
The position of a sample in a sodium iodide well counter has a dramatic effect on
the detection efficiency. When liquid samples in vials of a particular shape and size
are counted, the detection efficiency falls as the volume increases. Most nuclear
medicine in vitro tests require the comparison of a liquid sample from the patient
with a reference sample. It is crucial that both samples be in identical containers and
have identical volumes.
In addition, the high efficiency of the NaI well counter can cause unacceptable
dead-time count losses, even with sample activities in the microcurie range. It is
important to ensure that the activity placed in the well counter is sufficiently small
so that dead-time effects do not cause a falsely low count. In general, well counters
should not be used at apparent count rates exceeding about 5,000 cps, which limits
samples of 125I and 57Co to activities less than about 0.2 mCi. However, larger activities of some radionuclides may be counted without significant losses; for example,
activities of 51Cr as large as 5 mCi may be counted, because only about one out of
every ten decays yields a gamma ray.
Quality Assurance for the Sodium Iodide Thyroid

Probe and WellCounter
Both of these instruments should have energy calibrations (as discussed earlier for an
SCA system) performed daily, with the results recorded. A background count and a constancy test, using a source with a long half-life such as 137Cs, also should be performed
daily for both the well counter and the thyroid probe to test for radioactive contamination or instrument malfunction. On the day the constancy test is begun, a counting
window is set to tightly encompass the photopeak, and a count is taken and corrected
for background. Limits called action levels are established which, if exceeded, cause
the person performing the test to notify the chief technologist, physicist, or physician.
On each subsequent day, a count is taken using the same source, window setting, and
counting time; corrected for background; recorded; and compared with the action levels. If each days count were mistakenly compared with the previous days count instead
of the first days count, slow changes in the instrument would not be discovered. Periodically, a new first day count and action levels should be established, accounting for
decay of the radioactive source. This will prevent the constancy test count from exceeding an action level because of decreasing source activity. Also, spectra should be plotted
annually for commonly measured radionuclides, usually 123I and 131I for the thyroid
probe and perhaps 57Co (Schilling tests), 125I (radioimmunoassays), and 51Cr (red cell
volumes and survival studies) for the well counter, to verify that the SCA windows fit
the photopeaks. This testing is greatly simplified if the department has an MCA.
Dose Calibrator
A dose calibrator, shown in Figure 17-26, is used to measure the activities of dosages
of radiopharmaceuticals to be administered to patients. The U.S. Nuclear Regulatory
Commission (NRC) and state regulatory agencies require that dosages of x-ray and
gamma rayemitting radiopharmaceuticals be determined before administration to
patients. NRC allows administration of a unit dosage (an individual patient dosage
prepared by a commercial radiopharmacy, without any further manipulation of its
activity by the nuclear medicine department that administers it) without measurement. It also permits determination of activities of non-unit dosages by volumetric
measurements and mathematical calculations. Most dose calibrators are well-type
664
FIGURE 17-26 Dose calibrator. The

detector is a well-geometry ion chamber
filled with pressurized argon. The syringe
and vial holder (A) is used to place radioactive material, in this case in a syringe,
into the detector. This reduces exposure to
the hands and permits the activity to be
measured in a reproducible geometry. The
removable plastic insert (B) prevents contamination of the well.
ionization chambers that are filled with argon (Z 18) and pressurized to maximize
sensitivity. Most dose calibrators have shielding around their chambers to protect
users from the radioactive material being assayed and to prevent nearby sources of
radiation from affecting the measurements.
A dose calibrator cannot directly measure activity. Instead, it measures the intensity of the radiation emitted by a dosage of a radiopharmaceutical. The manufacturer
of a dose calibrator determines calibration factors relating the magnitude of the signal
from the detector to activity for specific radionuclides commonly used in nuclear
medicine. The user pushes a button or turns a dial on the dose calibrator to designate
the radionuclide being measured, thereby specifying a calibration factor, and the
dose calibrator displays the measured activity.
Operating Characteristics
Dose calibrators using ionization chambers are operated in current mode, thereby avoiding dead-time effects. They can accurately assay activities as large as 2 Ci. For the same
reasons, they are relatively insensitive and cannot accurately assay activities less than
about 1 Ci. In general, the identification and measurement of activities of radionuclides
in samples containing multiple radionuclides is not possible. The measurement accuracy
is affected by the position in the well of the dosages being measured, so it is important
that all measurements be made with the dosages at the same position. Most dose calibrators have large wells, which reduce the effect of position on the measurements.
Dose calibrators using large well-type ionization chambers are in general not
significantly affected by changes in the sample volume or container for most radionuclides. However, the measured activities of certain radionuclides, especially those
such as 111I, 123I, 125I, and 133Xe that emit weak x-rays or gamma rays, are highly
dependent on factors such as whether the containers (e.g., syringe or vial) are glass
665
or plastic and the thicknesses of the containers walls. There is currently no generally
accepted solution to this problem. Some radiopharmaceutical manufacturers provide
correction factors for these radionuclides; these correction factors are specific to the
radionuclide, the container, and the model of dose calibrator.
There are even greater problems with attenuation effects when assaying the dosages
of pure beta-emitting radionuclides, such as 32P and 89Sr. For this reason, the NRC does
not require unit dosages of these radionuclides to be assayed in a dose calibrator before
administration. However, most nuclear medicine departments assay these, but only to
verify that the activities as assayed by the vendor are not in error by a large amount.
Dose Calibrator Quality Assurance

Because the assay of activity using the dose calibrator is often the only assurance that the
patient is receiving the prescribed activity, quality assurance testing of the dose calibrator is required by the NRC and state regulatory agencies. The NRC requires the testing
to be in accordance with nationally recognized standards or the manufacturers instructions. The following set of tests is commonly performed to satisfy this requirement.
The device should be tested for accuracy upon installation and annually thereafter.
Two or more sealed radioactive sources (often 57Co and 137Cs) whose activities are
known within 5% are assayed. The measured activities must be within 10% of the
actual activities.
The device should also be tested for linearity (a measure of the effect that the
amount of activity has on the accuracy) upon installation and quarterly thereafter.
The most common method requires a vial of 99mTc containing the maximal activity
that would be administered to a patient. The vial is measured two or three times
daily until it decays to less than 30 Ci; the measured activities and times of the
measurements are recorded. One measurement is assumed to be correct and, from
this measurement, activities are calculated for the times of the other measurements.
No measurement may differ from the corresponding calculated activity by more than
10%. An alternative to the decay method for testing linearity is the use of commercially available lead cylindrical sleeves of different thickness that simulate radioactive
decay via attenuation (Figure 17-27). These devices must be calibrated prior to first
use by comparing their simulated decay with the decay method described above.
FIGURE 17-27 Set of cylindrical lead
sleeves used to test dose calibrator
linearity. These color-coded sleeves of
different thicknesses use attenuation to
simulate radioactive decay over periods
of 6, 12, 20, 30, 40, and 50 h for 99mTc.
Additional sleeves are available that can
extend the simulated decay interval up to
350 h. Photo courtesy of Fluke Biomedical
Radiation Management Services, Cleveland, Ohio.
666
The device should be tested for constancy before its first use each day. At least
one sealed source (usually 57Co) is assayed, and its measured activity, corrected for
decay, must not differ from its measured activity on the date of the last accuracy test
by more than 10%. (Most laboratories perform a daily accuracy test, which is more
rigorous, in lieu of a daily constancy test.)
Finally, the dose calibrator should be tested for geometry dependence on
installation. This is usually done by placing a small volume of a radiochemical,
often 99mTc pertechnetate, in a vial or syringe and assaying its activity after each of
several dilutions. If volume effects are found to affect measurements by more than
10%, correction factors must be determined. The geometry test must be performed
for syringe and vial sizes commonly used. Dose calibrators should also be appropriately tested after repair or adjustment.
The calibration factors of individual radionuclide settings should be verified
for radionuclides assayed for clinical purposes. This can be performed by placing
a source of any radionuclide in the dose calibrator, recording the indicated activity
using a clinical radionuclide setting (e.g., 131I), recording the indicated activity using
the setting of a radionuclide used for accuracy determination (e.g., 57Co or 137Cs),
and verifying that the ratio of the two indicated activities is that specified by the
manufacturer of the dose calibrator.
Molybdenum-99 Concentration Testing

When a 99Mo/99mTc generator is eluted, it is possible to obtain an abnormally large
amount of 99Mo in the eluate. (Radionuclide generators are discussed in Chapter16.)
If a radiopharmaceutical contaminated with 99Mo is administered to a patient, the
patient will receive an increased radiation dose (99Mo emits high-energy beta particles and has a 66-h half-life) and the quality of the resultant images may be degraded
by the high-energy 99Mo gamma rays. The NRC requires that any 99mTc to be administered to a human must not contain more than 0.15 kBq of 99Mo per MBq of 99mTc
(0.15 mCi of 99Mo per mCi of 99mTc) at the time of administration and that the first
elution of a generator must be assayed for 99Mo concentration.
The concentration of 99Mo is most commonly measured with a dose calibrator
and a special lead container that is supplied by the manufacturer. The walls of the
lead container are sufficiently thick to stop almost all of the gamma rays from 99mTc
(140 keV) but thin enough to be penetrated by many of the higher energy gamma
rays from 99Mo (740 and 778 keV). To perform the measurement, the empty lead
container is first assayed in the dose calibrator. Next, the vial of 99mTc is placed in
the lead container and assayed. Finally, the vial of 99mTc alone is assayed. The 99Mo
concentration is then obtained using the following equation:

Concentration K
Avial incontainer Aemptycontainer

Avial
[17-5]
where K is a correction factor supplied by the manufacturer of the dose calibrator

that accounts for the attenuation of the 99Mo gamma rays by the lead container.
Sr and 85Sr Concentration Testing
82
Myocardial perfusion can by assessed by PET using the radiopharmaceutical 82Rb

chloride. In clinical practice, 82Rb chloride is obtained from an 82Sr/82Rb generator. (These generators are discussed in Chapter 16.) When an 82Sr/82Rb generator is
eluted, it is possible to obtain an abnormally large amount of 82Sr or 85Sr, a radioactive
contaminant in the production of 82Sr, in the eluate. The NRC requires that any 82Rb
chloride to be administered to a human not contain more than 0.02 kBq of 82Sr or
667
0.2 kBq of 85Sr per mBq of 82Rb at the time of administration and that the concentrations of these radionuclides be measured before the first patient use each day. A dose
calibrator is generally used to assay the concentration of the unwanted contaminants
82
Sr and 85Sr in the eluate. The measurement procedures are contained in the package
insert provided by the manufacturer of the generators. The eluate is allowed to decay
for an hour so that only 82Sr and 85Sr remain. The sample is then assayed in the dose
calibrator, and the amounts of 82Sr and 85Sr are estimated from an assumed ratio.
17.7 Counting Statistics
Introduction
Sources of Error
There are three types of errors in measurements. The first is systematic error. Systematic
error occurs when measurements differ from the correct values in a systematic fashion.
For example, systematic error occurs in radiation measurements if a detector is used
in pulse mode at too high an interaction rate; dead-time count losses cause the measured count rate to be lower than the actual interaction rate. The second type of error
is random error. Random error is caused by random fluctuations in whatever is being
measured or in the measurement process itself. The third type of error is the blunder
(e.g., setting the single-channel analyzer window incorrectly for a single measurement).
Random Error in Radiation Detection

The processes by which radiation is emitted and interacts with matter are random
in nature. Whether a particular radioactive nucleus decays within a specified time
interval, the direction of an x-ray emitted by an electron striking the target of an x-ray
tube, whether a particular x ray passes through a patient to reach the film cassette
of an x-ray machine, and whether a gamma ray incident upon a scintillation camera
crystal is detected are all random phenomena. Therefore, all radiation measurements,
including medical imaging, are subject to random error. Counting statistics enable
judgments of the validity of measurements that are subject to random error.
Characterization of Data
Accuracy and Precision
If a measurement is close to the correct value, it is said to be accurate. If measurements are reproducible, they are said to be precise. Precision does not imply accuracy;
a set of measurements may be very close together (precise) but not close to the correct value (i.e., inaccurate). If a set of measurements differs from the correct value in
a systematic fashion (systematic error), the data are said to be biased.
Measures of Central TendencyMean and Median

Two measures of central tendency of a set of measurements are the mean (average)
and the median. The mean (x) of a set of measurements is defined as follows:

x1 x 2 x N
N
[17-6]
where N is the number of measurements.

To obtain the median of a set of measurements, they must first be put in order by
size. The median is the middle measurement if the number of measurements is odd,
and it is the average of the two midmost measurements if the number of measurements
668
iseven. For example, to obtain the median of the five measurements 8, 14, 5, 9, and12,
they are first sorted by size: 5, 8, 9, 12, and 14. The median is 9. The advantage of the
median over the mean is that the median is less affected by outliers. An outlier is a measurement that is much greater or much less than the others.
Measures of VariabilityVariance and Standard Deviation

The variance and standard deviation are measures of the variability (spread) of a set
of measurements. The variance (2) is determined from a set of measurements by
subtracting the mean from each measurement, squaring the differences, summing
the squares, and dividing by one less than the number of measurements:

( x1 x )2 ( x 2 x )2 ( x N x )2
N 1
[17-7]
where N is the total number of measurements and x is the sample mean.

The standard deviation () is the square root of the variance:
[17-8]
2
The fractional standard deviation (also referred to as the fractional error or coefficient
of variation) is the standard deviation divided by the mean:

Fractional standard deviation / x
[17-9]
Probability Distribution Functions for Binary Processes

Binary Processes
A trial is an event that may have more than one outcome. A binary process is a process in which a trial can only have two outcomes, one of which is arbitrarily called
a success. A toss of a coin is a binary process. The toss of a die can be considered a
binary process if, for example, a two is selected as a success and any other outcome
is considered to be a failure. Whether a particular radioactive nucleus decays during
a specified time interval is a binary process. Whether a particular x-ray or gamma
ray is detected by a radiation detector is a binary process. Table 17-2 lists examples
of binary processes.
A measurement consists of counting the number of successes from a specified
number of trials. Tossing ten coins and counting the number of heads is a measurement. Placing a radioactive sample on a detector and recording the number of events
detected is a measurement.
TABLE 17-2 BINARY PROCESSES
TRIAL
DEFINITION
OFA SUCCESS
PROBABILITY
OFA SUCCESS
Toss of a coin
Heads
1/2
Toss of a die
A four
1/6
Observation of a radioactive nucleus for a time t
It decays
1 et
Observation of a detector of efficiency E placed

near a radioactive nucleus for a time t
A count
E(1 et)
Source: Adapted from Knoll GF. Radiation detection and measurement, 4th ed. New York, NY: John Wiley, 2010.
669
Probability Distribution FunctionsBinomial,

Poisson, and Gaussian
A probability distribution function (pdf) describes the probability of obtaining each
outcome from a measurementfor example, the probability of obtaining six heads
in a throw of ten coins. There are three probability distribution functions relevant to
binary processesthe binomial, the Poisson, and the Gaussian (normal). The binomial distribution exactly describes the probability of each outcome from a measurement of a binary process:

P(x )
N!
p x (1 p )N x
x !( N x )!
[17-10]
where N is the total number of trials in a measurement, p is the probability of success

in a single trial, and x is the number of successes. The mathematical notation N!,
called factorial notation, is simply shorthand for the product
[17-11]

N! N (N 1) (N 2) 3 2 1
For example, 5! 54321 120. If we wish to know the probability of obtaining two heads in a toss of four coins, x 2, N 4, and p 0.5. The probability of
obtaining two heads is
P (two -heads )
4!
(0.5)2 (1 0.5)4 2 0.375
2!(4 2)!
Figure 17-28 is a graph of the binomial distribution. It can be shown that the
sum of the probabilities of all outcomes for the binomial distribution is 1.0 and that
the mean (x) and standard deviation () of the binomial distribution are as follows:
x pN and pN (1 p )
[17-12]
If the probability of a success in a trial is much less than 1 (not true for a toss of
a coin, but true for most radiation measurements), the standard deviation is approximated by the following:
pN (1 p )
pN x
[17-13]
Because of the factorials in Equation 17-10, it is difficult to use if either x or N is large.

The Poisson and Gaussian distributions are approximations to the binomial pdf that
are often used when x or N is large. Figure 17-29 shows a Gaussian distribution.
FIGURE 17-28 Binomial probability
distribution function when the probability of a success in a single trial (p) is
1/3 and the number of trials (N) is 10.
0.3
Probability
0.2
0.1
Number of Successes
10
670
FIGURE 17-29 Gaussian probability distribution function for pN 20.
0.09
0.08
0.07
Probability
0.06
0.05
0.04
0.03
0.02
0.01
0
10
15
20
25
30
Number of Successes
35
Estimating the Uncertainty of a Single Measurement

Estimated Standard Deviation
The standard deviation can be estimated, as previously described, by making several
measurements and applying Equations 17-7 and 17-8. Nevertheless, if the process
being measured is a binary process, the standard deviation can be estimated from a
single measurement. The single measurement is probably close to the mean. Because
the standard deviation is approximately the square root of the mean, it is also approximately the square root of the single measurement:
x
[17-14]
where x is the single measurement. The fractional standard deviation of a single measurement may also be estimated as:

Fractional error
x
1
x
x
x
[17-15]
For example, a single measurement of a radioactive source yields 1,256 counts.

The estimated standard deviation is
1,256 cts 35.4 cts
The fractional standard deviation is estimated as
Fractional error 35.4 cts/1,256 cts 0.028 2.8%
Table 17-3 lists the estimated fractional errors for various numbers of counts. The
fractional error decreases with the number of counts.
Confidence Intervals
Table 17-4 lists intervals about a measurement, called confidence intervals, for which the
probability of containing the true mean is specified. There is a 68.3% probability than the
true mean is within one standard deviation (1 ) of a measurement, a 95%probability
671
TABLE 17-3FRACTIONAL ERRORS (PERCENT
TANDARD DEVIATIONS) FOR SEVERAL

S
NUMBERS OFCOUNTS
COUNT
100
FRACTIONAL ERROR (%)
10.0
1,000
3.2
10,000
1.0
100,000
0.32
that it is within 2 of a measurement, and a 99.7% probability that it is within

3 of a measurement, assuming that the measurements follow a gaussian distribution.
For example, a count of 853 is obtained. Determine the interval about this count
in which there is a 95% probability of finding the true mean.
First, the standard deviation is estimated as
853 cts 29.2 cts
From Table 17-4, the 95% confidence interval is determined as follows:
853 cts 1.96 853 cts 1.96 (29.2 cts) 853 cts 57.2 cts
So, the 95% confidence interval ranges from 796 to 910 counts.
Propagation of Error
In nuclear medicine, calculations are frequently performed using numbers that incorporate random error. It is often necessary to estimate the uncertainty in the results of
these calculations. Although the standard deviation of a count may be estimated by
simply taking its square root, it is incorrect to calculate the standard deviation of the
result of a calculation by taking its square root. Instead, the standard deviations of the
actual counts must first be calculated and entered into propagation of error equations
to obtain the standard deviation of the result.
Multiplication or Division of a Number with Error

by a Number without Error
It is often necessary to multiply or divide a number containing random error by a
number that does not contain random error. For example, to calculate a count rate, a
TABLE 17-4 CONFIDENCE INTERVALS
INTERVAL ABOUT
MEASUREMENT
PROBABILITY THAT MEAN IS

WITHIN INTERVAL (%)
0.674
50.0
68.3
1.64
90.0
1.96
95.0
2.58
99.0
99.7
Source: Adapted from Knoll GF. Radiation detection and measurement, 4th
ed. New York, NY: John Wiley, 2010.
672
count (which incorporates random error) is divided by a counting time (which does
not involve significant random error). If a number x has a standard deviation and is
multiplied by a number c without random error, the standard deviation of the product cx is c. If a number x has a standard deviation and is divided by a number c
without random error, the standard deviation of the quotient x/c is /c.
For example, a 5-min count of a radioactive sample yields 952 counts. The count
rate is 952 cts/5 min 190 cts/min. The standard deviation and percent standard
deviation of the count are
952 cts 30.8 cts
Fractional error 30.8 cts/952 cts 0.032 3.2%
The standard deviation of the count rate is
30.8 cts/5 min 6.16 cts/min
Fractional error 6.16 cts/190 cts 3.2%
Notice that the percent standard deviation is not affected when a number is multiplied or divided by a number without random error.
Addition or Subtraction of Numbers with Error

It is often necessary to add or subtract numbers with random error. For example, a
background count may be subtracted from a count of a radioactive sample. Whether
two numbers are added or subtracted, the same equation is used to calculate the
standard deviation of the result, as shown in Table 17-5.
For example, a count of a radioactive sample yields 1,952 counts, and a background count with the sample removed from the detector yields 1,451 counts.
The count of the sample, corrected for background, is
1,952 cts 1,451 cts 501 cts
The standard deviation and percent standard deviation of the original sample
count are
s + b 1,952 cts 44.2 cts
Fractional error 44.2 cts/1,952 cts 2.3%
TABLE 17-5 PROPAGATION OF ERROR EQUATIONS

DESCRIPTION
OPERATION
STANDARD
DEVIATION
Multiplication of a number with random

error by a number without random error
cx
Division of a number with random error

by a number without random error
x/c
/c
Addition of two numbers containing

random errors
x1 x2
Subtraction of two numbers containing

random errors
x1 x2
12 22
12 22
Note: c is a number without random error, is the standard deviation of x, l is the

standard deviation of x1, and 2 is the standard deviation of x2.
673
The standard deviation and percent standard deviation of the background

count are
b 1, 451 cts 38.1 cts
Fractional error 38.1 cts/1,451 cts 2.6%
The standard deviation and percent standard deviation of the sample count corrected for background are
s (44.2 cts)2 (38.3cts)2 58.3 cts
Fractional error 58.1 cts/501 cts 11.6%
Note that the fractional error of the difference is much larger than the fractional
error of either count. The fractional error of the difference of two numbers of similar
magnitude can be much larger than the fractional errors of the two numbers.
Combination Problems
It is sometimes necessary to perform mathematical operations in series, for example,
to subtract two numbers and then to divide the difference by another number. The
standard deviation is calculated for each intermediate result in the series by entering
the standard deviations from the previous step into the appropriate propagation of
error equation in Table 17-5.
For example, a 5-min count of a radioactive sample yields 1,290 counts and a
5-min background count yields 451 counts. What is the count rate due to the sample
alone and its standard deviation?
First, the count is corrected for background:
Count 1,290 cts 451 cts 839 cts
The estimated standard deviation of each count and the difference are calculated:
sb 1,290 cts 35.9 cts and b 451 cts 21.2 cts
s (35.9 cts)2 (21.2 cts)2 41.7 cts
Finally, the count rate due to the source alone and its standard deviation are
calculated:
Count rate 839 cts/5 min 168 cts/min
s /c 41.7 cts/5 min 8.3 cts/min
SUGGESTED READING
Attix FH. Introduction to radiological physics and radiation dosimetry, New York, NY: John Wiley, 1986
Cherry, R., Simon et.al. Physics in Nuclear Medicine. 4th ed., Philadelphia: Saunders, 2011.
Knoll GF. Radiation detection and measurement, 4th ed. New York, NY: John Wiley, 2010.
Patton JA, Harris CC. Gamma well counter. In: Sandler MP, et al., eds. Diagnostic nuclear medicine, 3rd ed.
Baltimore, MD: Williams & Wilkins, 1996:5965.
Ranger NT. The AAPM/RSNA physics tutorial for residents: radiation detectors in nuclear medicine.
Radiographics 1999;19:481502.
Rzeszotarski MS. The AAPM/RSNA physics tutorial for residents: counting statistics. Radiographics
1999;19:765782.
Chapter
18
Nuclear ImagingThe
Scintillation Camera
Nuclear imaging produces images of the distributions of radionuclides in patients.
Because charged particles from radioactivity in a patient are almost entirely absorbed
within the patient, nuclear imaging uses gamma rays, characteristic x-rays (usually
from radionuclides that decay by electron capture), or annihilation photons (from
positron-emitting radionuclides) to form images.
To form a projection image, an imaging system must determine not only the
photon flux density (number of x- or gamma rays per unit area) at each point in
the image plane but also the directions of the detected photons. In x-ray transmission imaging, the primary photons travel known paths diverging radially from a
point (the focal spot of the x-ray tube). In contrast, the x- or gamma rays from the
radionuclide in each volume element of a patient are emitted isotropically (equally
in all directions). Nuclear medicine instruments designed to image gamma- and
x-rayemitting radionuclides use collimators that permit photons following certain
trajectories to reach the detector but absorb most of the rest. A heavy price is paid
for using collimationthe vast majority (typically well over 99.95%) of emitted photons is wasted. Thus, collimation, although necessary for the formation of projection
images, severely limits the performance of these devices. Instruments for imaging
positron (+)-emitting radionuclides can avoid collimation by exploiting the unique
properties of annihilation radiation to determine the directions of the photons.
The earliest widely successful nuclear medicine imaging device, the rectilinear scanner, which dominated nuclear imaging from the early 1950s through the late 1960s, used
a moving radiation detector to sample the photon fluence at a small region of the image
plane at a time1. This was improved upon by the use of a large-area position-sensitive
detector (a detector indicating the location of each interaction) to sample simultaneously
the photon fluence over the entire image plane. The Anger scintillation camera, which
currently dominates nuclear imaging, is an example of the latter method. The scanning
detector system is less expensive, but the position-sensitive detector system permits
more rapid image acquisition and has replaced single scanning detector systems.
Nuclear imaging devices using gas-filled detectors (such as multiwire proportional counters) have been developed. Unfortunately, the low densities of gases, even
when pressurized, yield low detection efficiencies for the x- and gamma-ray energies
commonly used in nuclear imaging. To obtain a sufficient number of interactions
to form statistically valid images without imparting an excessive radiation dose to
the patient, nearly all nuclear imaging devices in routine clinical use utilize solid
inorganic scintillators as detectors because of their superior detection efficiency.
Efforts are being made to develop nuclear imaging devices using semiconductor
detectors. This will require the development of a high atomic number, high-density
1
Cassen B, Curtis L, Reed A, Libby RL. Instrumentation for 1-131 use in medical studies. Nucleonics
1951: 9: 469.
674
Chapter 18 Nuclear ImagingThe Scintillation Camera
675
semiconductor detector of sufficient thickness for absorbing the x- and gamma rays
commonly used in nuclear imaging and that can be operated at room temperature.
The leading detector material to date is cadmium zinc telluride (CZT). A small field
of view (FOV) camera using CZT semiconductor detectors has been developed.
The attenuation of x-rays in the patient is useful in radiography, fluoroscopy, and
x-ray computed tomography and, in fact, is necessary for image formation. However,
in nuclear imaging, attenuation is usually a hindrance; it causes a loss of information
and, especially when it is very nonuniform, it is a source of artifacts.
The quality of a nuclear medicine image is determined not only by the performance
of the nuclear imaging device but also by the properties of the radiopharmaceutical
used. For example, the ability of a radiopharmaceutical to preferentially accumulate
in a lesion of interest largely determines the smallest such lesion that can be detected.
Furthermore, the dosimetric properties of a radiopharmaceutical determine the maximal activity that can be administered to a patient, which in turn affects the amount of
statistical noise (quantum mottle) in the image and the spatial resolution. For example,
in the 1950s and 1960s, radiopharmaceuticals labeled with I-131 and Hg-203 were
commonly used. The long half-lives of these radionuclides and their beta particle emissions limited administered activities to approximately 150 microcuries (Ci). These low
administered activities and the high-energy gamma rays of these radionuclides required
the use of collimators providing poor spatial resolution. Many modern radiopharmaceuticals are labeled with technetium-99m (Tc-99m), whose short half-life (6.01 h)
and isomeric transition decay (~88% of the energy is emitted as 140-keV gamma rays;
only ~12% is given to conversion electrons and other emissions unlikely to escape the
patient) permit activities of up to about 35 millicuries (mCi) to be administered. The
high gamma-ray flux from such an activity permits the use of high-resolution (lowefficiency) collimators. Radiopharmaceuticals are discussed in Chapter 16.
18.1 Planar Nuclear Imaging: The Anger Scintillation
Camera
The Anger gamma scintillation camera, developed by Hal O. Anger at the Donner
Laboratory in Berkeley, California, in the 1950s, is by far the most common nuclear
medicine imaging device.2 However, it did not begin to replace the rectilinear scanner
significantly until the late 1960s, when its spatial resolution became comparable to
that of the rectilinear scanner and Tc-99m-labeled radiopharmaceuticals, for which it
is ideally suited, became commonly used in nuclear medicine. Most of the advantages
of the scintillation camera over the rectilinear scanner stem from its ability simultaneously to collect data over a large area of the patient, rather than one small area at a
time. This permits the more rapid acquisition of images and enables dynamic studies
that depict the redistribution of radionuclides to be performed. Because the scintillation camera wastes fewer x- or gamma rays than earlier imaging devices, its images
have less quantum mottle (statistical noise) and it can be used with higher resolution
collimators, thereby producing images of better spatial resolution. The scintillation
camera is also more flexible in its positioning, permitting images to be obtained from
almost any angle. Although it can produce satisfactory images using x- or gamma rays
ranging in energy from about 70 keV (Tl-201) to 364 keV (I-131) or perhaps even 511
keV (F-18), the scintillation camera is best suited for imaging photons with energies
in the range of 100 to 200 keV. Figure 18-1 shows a modern scintillation camera.
Hal O. Anger, Scintillation Camera, Review of Scientific Instruments, 29, 1958, 2733.
676
FIGURE 18-1 Modern dual

rectangular head scintillation
camera. The two heads are in a
90 degree orientation for cardiac SPECT imaging (discussed in
Chapter 19). (Courtesy of Siemens
Medical Solutions.)
Scintillation cameras of other designs have been devised, and one, the multicrystal
scintillation camera (briefly described later in this chapter), achieved limited
commercial success. However, the superior performance of the Anger camera for most
applications has caused it to dominate nuclear imaging. The term scintillation camera
will refer exclusively to the Anger scintillation camera throughout this chapter.
Design and Principles of Operation

Detector and Electronic Circuits
A scintillation camera head (Fig. 18-2) contains a disk-shaped (mostly on older
cameras) or rectangular thallium-activated sodium iodide [NaI(Tl)] crystal, typically 0.95 cm ( 3 8 inch) thick, optically coupled to a large number (typically 37 to
91) of 5.1- to 7.6-cm (2- to 3-inch) diameter photomultiplier tubes (PMTs). PMTs
and NaI(Tl) scintillation crystals were described in Chapter 17. The NaI(Tl) crystals
of modern cameras have large areas; the rectangular crystals of one manufacturer
are 59 x 44.5 cm (23 x 17.4 in.), with a field-of-view of about 53 by 39 cm. Some
Analog voltage pulses to camera electronics
Pre
amp
Pre
amp
Pre
amp
Pre
amp
Pre
amp
Pre
amp
Pre
amp
Pre-amplifiers
PMT
PMT
PMT
PMT
PMT
PMT
PMT
Photomultiplier
tubes
Lucite light pipe

Nal(TI) crystal
collimator
FIGURE 18-2 Scintillation camera detector.
677
camera designs incorporate a Lucite light pipe between the glass cover of the crystal
and PMTs; in others, the PMTs are directly coupled to the glass cover. In most cameras, a preamplifier is connected to the output of each PMT. Between the patient and
the crystal is a collimator, usually made of lead, that only allows x- or gamma rays
approaching from certain directions to reach the crystal. The collimator is essential;
a scintillation camera without a collimator does not generate meaningful images.
Figure 18-2 shows a parallel-hole collimator.
The lead walls, called septa, between the holes in the collimator absorb most photons approaching the collimator from directions that are not aligned with the holes.
Most photons approaching the collimator from a nearly perpendicular direction pass
through the holes; many of these are absorbed in the sodium iodide crystal, causing
the emission of visible light and ultraviolet (UV) radiation. The light and UV photons
are converted into electrical signals and amplified by the PMTs. These signals are further amplified by the preamplifiers (preamps). The amplitude of the electrical pulse
produced by each PMT is proportional to the amount of light it received following
an x- or gamma-ray interaction in the crystal.
Because the collimator septa intercept most photons approaching the camera
along paths not aligned with the collimator holes, the pattern of photon interactions
in the crystal forms a two-dimensional projection of the three-dimensional activity
distribution in the patient. The PMTs closest to each photon interaction in the crystal
receive more light than those that are more distant, causing them to produce larger
voltage pulses. The relative amplitudes of the pulses from the PMTs following each
interaction contain sufficient information to determine the location of the interaction
in the plane of the crystal within a few mm.
Early scintillation cameras formed images on photographic film using only analog electronic circuitry. In the late 1970s, digital circuitry began to be used for some
functions. Modern scintillation cameras have an analog-to-digital converter (ADC) for
the signal from the preamplifier (preamp) following each PMT (Fig. 18-3). ADCs are
described in Chapter 5. The remaining circuits used for signal processing and image
formation are digital. The digitized signals from the preamps are sent to two circuits.
The position circuit receives the signals from the individual preamps after each x- or
gamma-ray interaction in the crystal and, by determining the centroid of these signals, produced an X-position signal and a Y-position signal that together specify the
position of the interaction in the plane of the crystal. The summing circuit adds the
signals from the individual preamps to produce an energy (Z) signal proportional
in amplitude to the total energy deposited in the crystal by the interaction. These
digital signals are then sent to correction circuits, described later in this chapter, to
correct position-dependent systematic errors in event position assignment and energy
determination (see Spatial Linearity and Uniformity, below). These correction circuits greatly improve the spatial linearity and uniformity of the images. The corrected
energy (Z) signal is sent to an energy discrimination circuit. An interaction in the cameras crystal is recorded as a count in the image only if the energy (Z) signal is within a
preset energy window. Scintillation cameras permit as many as four energy windows
to be set for imaging radionuclides, such as Ga-67 and In-111, which emit useful photons of more than one energy. Following energy discrimination, the X- and Y-position
signals are sent to a digital computer, where they are formed into a digital projection
image, as described in Section 18.2 below, that can be displayed on a monitor.
Collimators
The collimator of a scintillation camera forms the projection image by permitting xor gamma-ray photons approaching the camera from certain directions to reach the
678
ADCs
Pulses
from
individual
PMTs
Corrected
position
and
energy
signals
Position
and
energy
signals
Digital
position
circuit
Digital
Z
summing
(energy)
circuit
Digital
energy
and
spatial
linearity
correction
circuits
Digital
computer
Digital
Analog
FIGURE 18-3 Electronic circuits of a modern digital scintillation camera. An actual scintillation camera has
many more than four PMTs and ADCs.
c rystal while absorbing most of the other photons. Collimators are made of high atomic
number, high-density materials, usually lead. The most commonly used collimator is
the parallel-hole collimator, which contains thousands of parallel holes (Fig.18-4). The
holes may be round, square, or triangular; however, most state-of-the-art collimators
have hexagonal holes and are usually made from lead foil, although some are cast.
The partitions between the holes are called septa. The septa must be thick enough to
absorb most of the photons incident upon them. For this reason, collimators designed
Image in crystal
Image in crystal
Parallel
hole
Pinhole
Object
Object
Converging
Image in crystal
Diverging
Image in crystal
Object
Object
FIGURE 18-4 Collimators.
679
for use with radionuclides that emit higher energy photons have thicker septa. There
is an inherent compromise between the spatial resolution and efficiency (sensitivity) of collimators. Modifying a collimator to improve its spatial resolution (e.g., by
reducing the size of the holes or lengthening the collimator) reduces its efficiency.
Most scintillation cameras are provided with a selection of parallel-hole collimators.
These may include low-energy, high-sensitivity; low-energy, all-purpose (LEAP);
low-energy, high-resolution; medium-energy (suitable for Ga-67 and In-111),
high-energy (for I-131); and ultrahigh-energy (for F 18) collimators. The size
of the image produced by a parallel-hole collimator is not affected by the distance of
the object from the collimator. However, its spatial resolution degrades rapidly with
increasing collimator-to-object distance. The field-of-view (FOV) of a parallel-hole
collimator does not change with distance from the collimator.
A pinhole collimator (Fig. 18-4) is commonly used to produce magnified views
of small objects, such as the thyroid gland or a hip joint. It consists of a small (typically 3- to 5-mm diameter) hole in a piece of lead or tungsten mounted at the apex
of a leaded cone. Its function is identical to the pinhole in a pinhole photographic
camera. As shown in the figure, the pinhole collimator produces a magnified image
whose orientation is reversed. The magnification of the pinhole collimator decreases
as an object is moved away from the pinhole. If an object is as far from the pinhole as
the pinhole is from the crystal of the camera, the object is not magnified and, if the
object is moved yet farther from the pinhole, it is minified. (Clinical imaging is not
performed at these distances.) There are pitfalls in the use of pinhole collimators due
to the decreasing magnification with distance. For example, a thyroid nodule deep in
the mediastinum can appear to be in the thyroid itself. Pinhole collimators are used
extensively in pediatric nuclear medicine. On some pinhole collimators, the part
containing the hole can be removed and replaced with a part with a hole of another
diameter; this allows the hole size to be varied for different clinical applications.
A converging collimator (Fig. 18-4) has many holes, all aimed at a focal point in front
of the camera. As shown in the figure, the converging collimator magnifies the image.
The magnification increases as the object is moved away from the collimator. A disadvantage of a converging collimator is that its FOV decreases with distance from the collimator. A diverging collimator (Fig. 18-4) has many holes aimed at a focal point behind
the camera. It produces a minified image in which the amount of minification increases
as the object is moved away from the camera. A diverging collimator may be used to
image a part of a patient using a camera whose FOV, if a parallel-hole collimator were
used, would be smaller than the body part to be imaged. For example, a diverging collimator could enable a mobile scintillation camera with a 25- or 30-cm diameter crystal
to perform a lung study of a patient in the intensive care unit. If a diverging collimator
is reversed on a camera, it becomes a converging collimator. The diverging collimator
is seldom used today, because it has inferior imaging characteristics to the parallel-hole
collimator and the large rectangular crystals of most modern cameras render it unnecessary. The converging collimator is also seldom used; its imaging characteristics are
superior, in theory, to the parallel-hole collimator, but its decreasing FOV with distance
and varying magnification with distance have discouraged its use. However, a hybrid
of the parallel-hole and converging collimator, called a fan-beam collimator, may be
used in single photon emission computed tomography (SPECT) to take advantage of
the favorable imaging properties of the converging collimator (see Chapter 19).
Many special-purpose collimators, such as the seven-pinhole collimator, have
been developed. However, most of them have not enjoyed wide acceptance. The
performance characteristics of parallel-hole, pinhole, and fan-beam collimators are
discussed below (see Performance).
680
Principles of Image Formation

In nuclear imaging, which can also be called emission imaging, the photons from
each point in the patient are emitted isotropically. Figure 18-5 shows the fates of
the x- and gamma rays emitted in a patient being imaged. Some photons escape the
patient without interaction, some scatter within the patient before escaping, and
some are absorbed within the patient. Many of the photons escaping the patient
are not detected because they are emitted in directions away from the image receptor. The collimator absorbs the vast majority of those photons that reach it. As a
result, only a tiny fraction of the emitted photons (about 1 to 2 in 10,000 for typical
low-energy parallel-hole collimators) has trajectories permitting passage through the
collimator holes; thus, well over 99.9% of the photons emitted during imaging are
wasted. Some photons penetrate the septa of the collimator without interaction. Of
those reaching the crystal, some are absorbed in the crystal, some scatter from the
crystal, and some pass through the crystal without interaction. The relative probabilities of these events depend on the energies of the photons and the thickness of the
crystal. Of those photons absorbed in the crystal, some are absorbed by a single photoelectric absorption, whereas others undergo one or more Compton scatters before
a photoelectric absorption. It is also possible for two photons to interact simultaneously with the crystal; if the energy (Z) signal from the coincident interactions is
within the energy window of the energy discrimination circuit, the result will be a
single count mispositioned in the image. The fraction of simultaneous interactions
increases with the interaction rate of the camera.
Collimator septal penetration

Crystal
Scatter in Patient
Crystal
Crystal
Coincident interactions
Crystal
FIGURE 18-5 Ways that x- and gamma rays interact with a scintillation camera. All of these, other than
the ones depicted in the upper left, cause a loss of contrast and spatial resolution and add statistical noise.
However, interactions by photons that have scattered though large angles and many coincident interactions
are rejected by pulse height discrimination circuits.
681
Interactions in the crystal of photons that have been scattered in the patient,
hotons that have penetrated the collimator septa, photons that undergo one or
p
more scatters in the crystal, and coincident interactions reduce the spatial resolution
and image contrast and increase random noise. The function of the cameras energy
discrimination circuits (also known as pulse height analyzers) is to reduce this loss
of resolution and contrast by rejecting photons that scatter in the patient or result in
coincident interactions. Unfortunately, the limited energy resolution of the camera
causes a wide photopeak, and low-energy photons can scatter through large angles
with only a small energy loss. For example, a 140-keV photon scattering 45 degrees
will only lose 7.4% of its energy. An energy window that encompasses most of the
photopeak will unfortunately still accept a significant fraction of the scattered photons and coincident interactions.
It is instructive to compare single photon emission imaging with x-ray transmission imaging (Table 18-1). In x-ray transmission imaging, including radiography and fluoroscopy, an image is projected on the image receptor because
the x-rays originate from a very small source that approximates a point. In comparison, the photons in nuclear imaging are emitted isotropically throughout the
patient and therefore collimation is necessary to form a projection image. Furthermore, in x-ray transmission imaging, x-rays that have scattered in the patient can
be distinguished from primary x-rays by their directions and thus can be largely
removed by grids. In emission imaging, primary photons cannot be distinguished
from scattered photons by their directions. The collimator removes about the same
fraction of scattered photons as it does primary photons and, unlike the grid in
x-ray transmission imaging, does not reduce the fraction of counts in the resultant
image due to scatter. Scattered photons in nuclear imaging can only be differentiated from primary photons by energy, because scattering reduces photon energy.
In emission imaging, energy discrimination is used to reduce the fraction of counts
in the image caused by scattered radiation. Finally, nuclear imaging devices must
use pulse mode (the signal from each interaction is processed separately) for event
localization and so that interaction-by-interaction energy discrimination can be
employed; x-ray transmission imaging systems have photon fluxes that are too
high to allow a radiation detector to detect, discriminate, and record on a photonby-photon basis.
Alternative Camera Designthe Multielement Camera

An alternative to the Anger scintillation camera for nuclear medicine imaging is the multidetector element camera. An image receptor of such a camera is a two-dimensional array
of many small independent detector elements. Such cameras are used with collimators
for image formation, as are Anger scintillation cameras.
TABLE 18-1COMPARISON OF SINGLE-PHOTON NUCLEAR IMAGING
WITH X-RAY TRANSMISSION IMAGING
X-ray transmission
imaging
PRINCIPLE OF IMAGE
FORMATION
SCATTER REJECTION
PULSE OR CURRENT MODE
Point source
Current
Scintillation camera Collimation
Grid or air gap
Pulse height discrimination Pulse
682
One such camera, the Baird Atomic System 77, achieved limited clinical
a cceptance in the 1970s and early 1980s. It consisted of 294 NaI(Tl) scintillation
crystals arranged in an array of 14 rows and 21 columns. A clever arrangement of
light pipes was used to route the light from the crystals to 35 PMTs, with all crystals
in a row connected to one PMT and all crystals in a column being connected to
another PMT. Light signals being simultaneously received by the PMT for a column
and another PMT for a row identified the crystal in which a gamma ray or x-ray
interaction occurred. Since that time, other multielement cameras have been developed. Each detector element of one design is a CsI(Tl) scintillation crystal optically
coupled to a single photodiode and, in another design, each detector element is a
CZT semiconductor detector.
The intrinsic spatial resolution of a multi-element detector camera is determined
by the physical dimensions of the independent detector elements. An advantage to
the multielement camera design is that very high interaction rates can be tolerated
with only limited dead-time losses because of the independent detectors in the image
receptor. A disadvantage is the complexity of the electronics needed to process the
signals from such a large number of independent detectors.
Performance
Measures of Performance
Measures of the performance of a scintillation camera with the collimator attached
are called system or extrinsic measurements. Measures of camera performance with
the collimator removed are called intrinsic measurements. System measurements give
the best indication of clinical performance, but intrinsic measurements are often
more useful for comparing the performance of different cameras, because they isolate
camera performance from collimator performance.
Uniformity is a measure of a cameras response to uniform irradiation of the detector
surface. The ideal response is a perfectly uniform image. Intrinsic uniformity is usually
measured by placing a point radionuclide source (typically 150 Ci of Tc-99m) in front
of the uncollimated camera. The source should be placed at least four times the largest
dimension of the crystal away to ensure uniform irradiation of the camera surface and
at least five times away, if the uniformity image is to be analyzed quantitatively using
a computer. System uniformity, which reveals collimator as well as camera defects, is
assessed by placing a uniform planar radionuclide source in front of the collimated
camera. Solid planar sealed sources of Co-57 (typically 5 to 10 mCi when new) and
planar sources that may be filled with a Tc-99m solution are commercially available.
A planar source should be large enough to cover the crystal of the camera. The uniformity of the resultant images may be analyzed by a computer or evaluated visually.
Spatial resolution is a measure of a cameras ability to accurately portray spatial variations in activity concentration and to distinguish as separate radioactive
objects in close proximity. The system spatial resolution is evaluated by acquiring an
image of a line source, such as a capillary tube filled with Tc-99m, using a computer interfaced to the collimated camera and determining the line spread function
(LSF). The LSF, described in Chapter 4, is a cross-sectional profile of the image of
a line source. The full-width-at-half-maximum (FWHM), the full-width-at-tenthmaximum, and the modulation transfer function (described in Chapter 4) may all
be derived from the LSF.
The system spatial resolution, if measured in air so that scatter is not present, is
determined by the collimator resolution and the intrinsic resolution of the camera.
683

Projected radiation
profile
FIGURE 18-6 Collimator resolution.
FWHM
Line source
(seen end-on)
The collimator resolution is defined as the FWHM of the radiation transmitted through
the collimator from a line source (Fig. 18-6). It is not directly measured, but calculated from the system and intrinsic resolutions. The intrinsic resolution is determined
quantitatively by acquiring an image with a sheet of lead containing one or more thin
slits placed against the uncollimated camera using a point source. The point source
should be positioned at least five times the largest dimension of the cameras crystal
away. The system (RS), collimator (RC), and intrinsic (RI) resolutions, as indicated by
the FWHMs of the LSFs, are related by the following equation:

RS RC2 RI2
[18-1]
Some types of collimators magnify (converging, pinhole) or minify (diverging) the

image. For these collimators, the system and collimator resolutions should be corrected for the collimator magnification, so that they refer to distances in the object
rather than distances in the camera crystal:

RS RC 2 (RI / m )2
[18-2]
where RS' = RS/m is the system resolution corrected for magnification, RC RC/m is
the collimator resolution corrected for magnification, and m is the collimator magnification (image size in crystal/object size). The collimator magnification is determined as follows:
m 1.0 for parallel-hole collimators
m (pinhole-to-crystal distance)/(object-to-pinhole distance) for pinhole collimators
m f/(f x) for converging collimators
m f/(f + x) for diverging collimators
where f is the distance from the crystal to the focal point of the collimator and x is the
distance of the object from the crystal. It will be seen later in this section that the collimator and system resolutions degrade (FWHM of the LSF, corrected for collimator magnification, increases) with increasing distance between the line source and collimator.
684
As can be seen from Equation 18-2, collimator magnification reduces the deleterious effect of intrinsic spatial resolution on the overall system resolution. Geometric
magnification in x-ray transmission imaging reduces the effect of image receptor blur
for the same reason.
In routine practice, the intrinsic spatial resolution is semiquantitatively evaluated by imaging a parallel line or a four-quadrant bar phantom (Fig. 18-7) in contact
with the camera face, using a planar radionuclide source if the camera is collimated
or a distant point source if the collimator is removed, and visually noting the smallest bars that are resolved. By convention, the widths of the bars are the same as the
widths of the spaces between the bars. The size of the smallest bars that are resolved
is approximately related to the FWHM of the LSF:

(FWHM of the LSF) 1.7 (Size of smallest bars resolved)
[18-3]
FIGURE 18-7 Use of a bar phantom for evaluating spatial resolution. A plastic rectangular flood source
containing up to10 mCi (370 MBq) of uniformly distributed Co-57 (A) is placed on top of the bar phantom
(B), which is resting on top of the camera head collimator. The bar phantom has four quadrants, each consisting of parallel lead bars with a specific width and spacing between the bars. An image (C) is acquired with
approximately 10 million counts and the lines are visually inspected for linearity and determination of closest
spacing between the bars that can be seen resolved as separate lines. A typical four-quadrant bar phantom for
a modern scintillation camera might have 2.0-, 2.5-, 3.0-, and 3.5-mm wide lead bars, with the widths of the
spaces equal to the widths of the bars. This extrinsic resolution test (performed with the collimator attached
to the camera) evaluates the system resolution. An intrinsic resolution test is performed with the collimator
removed from the camera head. In this case, the bar phantom is placed against the crystal of the uncollimated
camera head and irradiated by a point source containing 100 to 200 Ci (3.7 to 7.4 MBq) of Tc-99m in a
syringe placed at least four camera crystal diameters away.
685
Spatial linearity (lack of spatial distortion) is a measure of the cameras ability to

portray the shapes of objects accurately. It is determined from the images of a bar
phantom, line source, or other phantom by assessing the straightness of the lines in
the image. Spatial nonlinearities can significantly degrade the uniformity, as will be
discussed later in this chapter.
Multienergy spatial registration, commonly called multiple window spatial registration, is a measure of the cameras ability to maintain the same image magnification,
regardless of the energies deposited in the crystal by the incident x- or gamma rays.
(Image magnification is defined as the distance between two points in the image
divided by the actual distance between the two points in the object being imaged.)
Higher energy x- and gamma-ray photons produce larger signals from the individual
PMTs than do lower energy photons. The position circuit of the scintillation camera
normalizes the X- and Y-position signals by the energy signal, so that the position
signals are independent of the deposited energy. If a radionuclide emitting useful
photons of several energies, such as Ga-67 (93-, 185-, and 300-keV gamma rays),
is imaged and the normalization is not properly performed, the resultant image will
be a superposition of images of different magnifications (Fig. 18-8). The multienergy
spatial registration can be tested by imaging several point sources of Ga-67, offset
from the center of the camera, using only one major gamma ray at a time. The centroid of the count distribution of each source should be at the same position in the
image for all three gamma-ray energies.
The system efficiency (sensitivity) of a scintillation camera is the fraction of x- or
gamma rays emitted by a source that produces counts in the image. The system
efficiency is important because it, in conjunction with imaging time, determines the
amount of quantum mottle (graininess) in the images. The system efficiency (Es) is
the product of three factors: the collimator efficiency (Ec), the intrinsic efficiency of
the crystal (Ei), and the fraction (f) of interacting photons accepted by the energy
discrimination circuits:

Es Ec Ei f
[18-4]
The collimator efficiency is the fraction of photons emitted by a source that penetrate
the collimator holes. In general, it is a function of the distance between the source
and the collimator and the design of the collimator. The intrinsic efficiency, the fraction
FIGURE 18-8 Multienergy spatial registration. Left. A simulated image of two point sources of a radionuclide emitting gamma rays of three different energies, illustrating proper normalization of the X- and Y-position signals for deposited energy. Right. A simulated image of the same two point sources, showing improper
adjustment of the energy normalization circuit. The maladjustment causes higher energy photons to produce
larger X and Y values than lower energy photons interacting at the same position in the cameras crystal,
resulting in multiple images of each point source. A less severe maladjustment would cause each point source
to appear as an ellipse, rather than three discrete dots.
686
of photons penetrating the collimator that interact with the NaI(Tl) crystal, is determined by the thickness of the crystal and the energy of the photons:

E i 1 e x
[18-5]
where is the linear attenuation coefficient of sodium iodide and x is the thickness of
the crystal. The last two factors in Equation 18-4 can be combined to form the photopeak efficiency, defined as the fraction of photons reaching the crystal that produce
counts in the photopeak of the energy spectrum:

Ep Ei f
[18-6]
This equation assumes that the window of the energy discrimination circuit has been
adjusted to encompass the photopeak exactly. In theory, the system efficiency and
each of its components range from zero to one. In reality, low-energy all-purpose
parallel-hole collimators have efficiencies of about 2 104 and low-energy highresolution parallel-hole collimators have efficiencies of about 1 104.
The energy resolution of a scintillation camera is a measure of its ability to distinguish between interactions depositing different energies in its crystal. A camera
with superior energy resolution is able to reject a larger fraction of photons that
have scattered in the patient or have undergone coincident interactions, thereby
producing images of better contrast and less relative random noise. The energy
resolution is measured by exposing the camera to a point source of a radionuclide,
usually Tc-99m, emitting monoenergetic photons and acquiring a spectrum of the
energy (Z) pulses, using either the nuclear medicine computer interfaced to the
camera, if it has the capability, or a multichannel analyzer. The energy resolution
is calculated from the FWHM of the photopeak (see Pulse Height Spectroscopy
in Chapter 17). The FWHM is divided by the energy of the photon (140 keV
for Tc-99m) and is expressed as a percentage. A wider FWHM implies poorer energy
resolution.
Scintillation cameras are operated in pulse mode (see Design and Principles of
Operation, above) and therefore suffer from dead-time count losses at high interaction rates. They behave as paralyzable systems (see Chapter 17); the indicated count
rate initially increases as the imaged activity increases, but ultimately reaches a maximum and decreases thereafter. The count-rate performance of a camera is usually
specified by (1) the observed count rate at 20% count loss and (2) the maximal count
rate. These count rates may be measured with or without scatter. Both are reduced
when measured with scatter. Table 18-2 lists typical values for modern cameras.
These high count rates are usually achieved by implementing a high count-rate mode
that degrades the spatial and energy resolutions.
Some scintillation cameras can correctly process the PMT signals when two or
more interactions occur simultaneously in the crystal, provided that the interactions
are separated by sufficient distance. This significantly increases the number of interactions that are correctly registered at high interaction rates.
Design Factors Determining Performance

Intrinsic Spatial Resolution and Intrinsic Efficiency
The intrinsic spatial resolution of a scintillation camera is determined by the types
of interactions by which the x- or gamma rays deposit energy in the cameras crystal and the statistics of the detection of the visible light photons emitted following
these interactions. The most important of these is the random error associated
687
TABLE 18-2TYPICAL INTRINSIC PERFORMANCE CHARACTERISTICS OF A
MODERN SCINTILLATION CAMERA, MEASURED BY NEMA

PROTOCOL
Intrinsic spatial resolution (FWHM of LSF for 140 keV)a
2.74.2 mm
Energy resolution (FWHM of photopeak for 140 keV photons)
9.2%11%
Integral uniformity (max. pixel min. pixel)/(max. pixel

+ min. pixel)
2%5%
Absolute spatial linearity
Less than 1.5 mm
Observed count rate at 20% count loss (measured

without scatter)
110,000260,000 counts/s
Observed maximal count rate (measured without scatter)
170,000500,000 counts/s
Intrinsic spatial resolution is for a 0.95-cm (3/8-inch) thick crystal; thicker crystals cause slightly worse
spatial resolution.
FWHM, full width at half maximum; LSF, line spread function; NEMA, National Electrical Manufacturers
Association.
with the collection of ultraviolet (UV) and visible light photons and subsequent
production of electrical signals by the PMTs. Approximately one UV or visible
light photon is emitted in the NaI crystal for every 25 eV deposited by an x- or
gamma ray. For example, when a 140-keV gamma ray is absorbed by the crystal,
approximately 140,000/25 5,600 UV or visible light photons are produced.
About two-thirds of these, approximately 3,700 photons, reach the photocathodes of the PMTs. Only about one out of every five of these causes an electron
to escape a photocathode, giving rise to about 750 electrons. These electrons are
divided among the PMTs, with the most being produced in the PMTs closest to
the interaction. Thus, only a small number of photoelectrons is generated in any
one PMT after an interaction. Because the processes by which the absorption of a
gamma ray causes the release of electrons from a photocathode are random, the
pulses from the PMTs after an interaction contain significant random errors that,
in turn, cause errors in the X and Y signals produced by the position circuit of
the camera and the energy (Z) signal. These random errors limit both the intrinsic spatial resolution and the energy resolution of the camera. The energy of the
incident x- or gamma rays determines the amount of random error in the event
localization process; higher energy photons provide lower relative random errors
and therefore superior intrinsic spatial resolution. For example, the gamma rays
from Tc-99m (140 keV) produce better spatial resolution than do the x-rays from
Tl-201 (69 to 80 keV), because each Tc-99m gamma ray produces about twice
as many light photons when absorbed. There is relatively little improvement in
intrinsic spatial resolution for gamma-ray energies above 250 keV because the
improvement in spatial resolution due to more visible light photons is largely
offset by the increased likelihood of scattering in the crystal before photoelectric
absorption (discussed below).
The quantum detection efficiency of the PMTs in detecting the UV and visible
light photons produced in the crystal is the most significant factor limiting the intrinsic spatial resolution (typically only 20% to 25% of the UV and visible light photons
incident on a photocathode contribute to the signal from the PMT). The size of
the PMTs also affects the spatial resolution slightly; using PMTs of smaller diameter
improves the spatial resolution by providing better sampling of the light emitted
following each interaction in the crystal.
688
A thinner NaI crystal provides better intrinsic spatial resolution than a thicker
crystal. A thinner crystal permits less spreading of the light before it reaches the
PMTs. Furthermore, a thinner crystal reduces the likelihood of an incident x- or
gamma ray undergoing one or more Compton scatters in the crystal followed by
photoelectric absorption. Compton scattering in the crystal can cause the centroid
of the energy deposition to be significantly offset from the site of the initial interaction in the crystal. The likelihood of one or more scatters in the crystal preceding the
photoelectric absorption increases with the energy of the x- or gamma ray.
The intrinsic efficiency of a scintillation camera is determined by the thickness of
the crystal and the energy of the incident x- or gamma rays. Figure 18-9 is a graph of
photopeak efficiency (fraction of incident x- or gamma rays producing counts in the
photopeak) as a function of photon energy for NaI(Tl) crystals 0.635 cm (inch),
1.27 cm ( inch), and 2.54 cm (1 inch) thick. Most modern cameras have 0.95-cm
( 3 8 inch) thick crystals. The photopeak efficiency of these crystals is greater than
80% for the 140-keV gamma rays from Tc-99m, but less than 30% for the 364-keV
gamma rays from I-131. Some cameras, designed primarily for imaging radionuclides such as Tl-201 and Tc-99m, which emit low-energy photons, have 0.635-cm
( inch) thick crystals. Other cameras, designed to provide greater intrinsic efficiency for imaging radionuclides such as I-131 which emit high-energy gamma rays,
have 1.27- to 2.5-cm ( to 1 inch) thick crystals.
There is a design compromise between the intrinsic efficiency of the camera, which
increases with crystal thickness, and intrinsic spatial resolution, which degrades with crystal
thickness. This design compromise is similar to the design compromise between the
spatial resolution of scintillator-based image receptors used in radiography and fluoroscopy, which deteriorates with increasing phosphor layer thickness, and detection
efficiency, which improves with increasing thickness.
Collimator Resolution and Collimator Efficiency

The collimator spatial resolution, as defined above, of multihole collimators (parallel-hole, converging, and diverging) is determined by the geometry of the holes.
The spatial resolution improves (narrower FWHM of the LSF) as the diameters of
the holes are reduced and the lengths of the holes (thickness of the collimator) are
increased. Unfortunately, changing the hole geometry to improve the spatial resolution in general reduces the collimators efficiency. The resultant compromise between
collimator efficiency and collimator resolution is the single most significant limitation on
scintillation camera performance.
1
Photopeak efficiency
FIGURE 18-9 Calculated photopeak efficiency as a function

of x- or gamma-ray energy for
NaI(Tl) crystals. (Data from Anger
HO, Davis DH. Gamma-ray detection efficiency and image resolution in sodium iodide. Rev Sci Instr
1964;35:693.)
1.0 inch thick
0.5
0.3
0.5 inch thick
0.2
0.25 inch thick
0.1
0.05
100
200
300
400
Gamma ray energy (keV)
500
689
The spatial resolution of a parallel-hole collimator decreases (i.e., FWHM of the

LSF increases) linearly as the collimator-to-object distance increases. This degradation of collimator spatial resolution with increasing collimator-to-object distance is
also one of the most important factors limiting scintillation camera performance.
However, the efficiency of a parallel-hole collimator is nearly constant over the collimator-to-object distances used for clinical imaging. Although the number of photons
passing through a particular collimator hole decreases as the square of the distance,
the number of holes through which photons can pass increases as the square of the
distance. The efficiency of a parallel-hole collimator, neglecting septal penetration,
is approximately

EC
A
g
4 l2
[18-7]
where A is the cross-sectional area of a single collimator hole, l is the length of a hole
(i.e., the thickness of the collimator), and g is the fraction of the frontal area of the
collimator that is not blocked by the collimator septa (g = total area of holes in collimator face/area of collimator face).
Figure 18-10 depicts the line spread function (LSF) of a parallel-hole collimator
as a function of source-to-collimator distance. The width of the LSF increases with
distance. Nevertheless, the area under the LSF (total number of counts) does not
significantly decrease with distance.
The spatial resolution of a pinhole collimator, along its central axis, corrected for
collimator magnification, is approximately equal to

RC d
f x

f
[18-8]
Collimator
line spread function
(projected radiation profile)
Collimator
Line source
FIGURE 18-10 Line spread function (LSF) of a parallel-hole collimator as a function of source-to-collimator
distance. The full-width-at-half-maximum (FWHM) of the LSF increases linearly with distance from the source
to the collimator; however, the total area under the LSF (photon fluence through the collimator) decreases very
little with source to collimator distance. (In both figures, the line source is seen end-on.)
690
TABLE 18-3THE EFFECT OF INCREASING COLLIMATOR-TO-OBJECT
DISTANCE ON COLLIMATOR PERFORMANCE PARAMETERS

COLLIMATOR SPATIAL RESOLUTIONa EFFICIENCY
FIELD SIZE MAGNIFICATION
Parallel hole Decreases
Approximately Constant
constant
Constant (m = 1.0)
Converging
Decreases
Increases
Decreases Increases (m >1 at collimator

surface)
Diverging
Decreases
Decreases
Increases
Decreases (m <1 at collimator

surface)
Pinhole
Decreases
Decreases
Increases
Decreases (m largest near

pinhole)
Spatial resolution corrected for magnification.
where d is the diameter of the pinhole, f is the distance from the crystal to the pinhole, and x is the distance from the pinhole to the object. The efficiency of a pinhole
collimator, along its central axis, neglecting penetration around the pinhole, is

E C d 2/16x 2
[18-9]
Thus, the spatial resolution of the collimator improves (i.e., RC decreases) as the
diameter of the pinhole is reduced, but the collimator efficiency decreases as the
square of the pinhole diameter. Both the collimator spatial resolution and the efficiency are best for objects close to the pinhole (small x). The efficiency decreases
rapidly with increasing pinhole-to-object distance.
The spatial resolution of all collimators decreases with increasing gamma-ray
energy because of increasing collimator septal penetration or, in the case of pinhole
collimators, penetration around the pinhole. Table 18-3 compares the characteristics
of different types of collimators.
System Spatial Resolution and Efficiency

The system spatial resolution and efficiency are determined by the intrinsic and
collimator resolutions and efficiencies, as described by Equations 18-2 and 18-4.
Figure18-11 (top) shows the effect of object-to-collimator distance on system spatial resolution. The system resolutions shown in this figure are corrected for magnification. The system spatial resolution, when corrected for magnification, is degraded
(i.e., FWHM of the LSF increases) as the collimator-to-object distance increases for
all types of collimators. This degradation of resolution with increasing patient-tocollimator distance is among the most important factors in image acquisition.
Technologists should make every effort to minimize this distance during clinical
imaging.
Figure 18-11 (bottom) shows the effect of object-to-collimator distance on system efficiency. The system efficiency with a parallel-hole collimator is nearly constant
with distance. The system efficiency with a pinhole collimator decreases significantly
with distance. The system efficiency of a fan-beam collimator (described in the following chapter) increases with collimator-to-object distance.
Modern scintillation cameras with 0.95-cm ( 3 8 inch) thick crystals typically have
intrinsic spatial resolutions slightly less than 4.0 mm FWHM for the 140-key gamma
rays of Tc-99m. One manufacturers low-energy high-resolution parallel-hole collimator has a resolution of 1.5 mm FWHM at the face of the collimator and 6.4 mm at
System resolution (mm FWHM)
691
LEAP
18
16
LEHR
14
12
Fanbeam
Pinhole
LEUHR
10
8
6
4
10
15
20
25
30
Object-to-collimator distance (cm)
System efficiency
1.5
LEAP
Fanbeam
LEHR
0.5
LEUHR
Pinhole
10
15
20
Object-to-collimator distance
25
30
FIGURE 18-11 System spatial resolution (top) and efficiency (bottom) as a function of object-to-collimator
distance (in cm). System resolutions for pinhole and fan-beam collimators are corrected for magnification.
System efficiencies are relative to a low-energy, all-purpose (LEAP) parallel-hole collimator (system efficiency
340 cpm/Ci Tc-99m for a 0.95-cm- thick crystal). LEHR, low-energy, high-resolution parallel-hole collimator;
LEUHR, low-energy, ultra-high-resolution parallel-hole collimator. (Data courtesy of the late William Guth of
Siemens Medical Systems, Nuclear Medicine Group.)
10 cm from the collimator. The system resolutions, calculated from Equation 18-1,
are as follows:

RS (1.5 mm)2 (4.0 mm)2 4.3 mm at 0 cm
[18-9a]
RS (6.4 mm)2 (4.0 mm)2 7.5 mm at 10 cm
[18-9b]
and

Thus, the system resolution is only slightly worse than the intrinsic resolution at the
collimator face, but is largely determined by the collimator resolution at typical imaging distances for internal organs.
Spatial Linearity and Uniformity

Fully analog scintillation cameras, now obsolete, suffered from significant spatial
nonlinearity and nonuniformity. These distortions were caused by systematic (nonrandom) effects, which vary with position on the face of the camera, in the detection
692
FIGURE 18-12 Pairs of uniformity images, lead slit-mask (lead sheet with thin slits) images, and orthogonal
hole phantom (lead sheet with a rectangular array of holes) images, with scintillation cameras digital correction
circuitry disabled (top) to demonstrate nonuniformities and spatial nonlinearities inherent to a scintillation camera
and with correction circuitry functioning (bottom), demonstrating effectiveness of linearity and energy (Z) signal
correction circuitry. (Photographs courtesy of Everett W. Stoub, Ph.D., formerly of Siemens Gammasonics, Inc.)
and collection of signals from individual x- or gamma rays. Because these positiondependent effects are not random, corrections for them can be made. Modern digital
cameras use tables of position-specific correction factors for this function.
Spatial nonlinearity (Fig. 18-12 top center and top right) is caused by the nonrandom (i.e., systematic) mispositioning of events. It is mainly due to the calculated
locations of the interactions being shifted in the resultant image toward the center of
the nearest PMT by the position circuit of the camera. Modern cameras have digital
circuits that use tables of correction factors to correct each pair of X- and Y-position
signals for spatial nonlinearities (Fig. 18-13). One lookup table contains an X-position correction for each portion of the crystal and another table contains corrections
for the Y direction. Each pair of digital position signals is used to look up a pair of X
and Y corrections in the tables. The corrections are added to the uncorrected X and Y
values. The corrected X and Y values (X and Y in Fig. 18-13) are sent to a computer
or other device that accepts them in digital form for image formation.
FIGURE 18-13 Spatial linearity correction circuitry of a
scintillation camera.
X and Y correction
lookup tables
X
Y
Digital
position
signals
X' = X + X
X
Y
Corrected
position
signals
X'
Y' = Y + Y
Corrections added to
digital X and Y values
Y'

Areas of decreased
count density
Photomultiplier
tubes
693
FIGURE 18-14 Spatial nonlinearities cause nonuniformities. The two vertical wavy lines
represent straight lines in the
object that have been distorted.
The scintillation cameras position circuit causes the locations
of individual counts to be shifted
toward the center of the nearest photomultiplier tube (PMT),
causing an enhanced count density toward the center of the PMT
and a decreased count density
between the PMTs, as seen in the
top left image in Figure 18-12.
Areas of increased
count density
There are three major causes of nonuniformity. The first is spatial nonlinearities.
As shown in Figure 18-14, the systematic mispositioning of events imposes local
variations in the count density. Spatial nonlinearities that are almost imperceptible
can cause significant nonuniformities. The linearity correction circuitry previously
described effectively corrects this source of nonuniformity.
The second major cause of nonuniformity is that the position of the interaction
in the crystal affects the magnitude of the energy (Z) signal. It may be caused by local
variations in the crystal in the light generation and light transmission to the PMTs
and by variations in the light detection and gains of the PMTs. If these regional variations in energy signal are not corrected, the fraction of interactions rejected by the
energy discrimination circuits will vary with position in the crystal. These positional
variations in the magnitude of the energy signal are corrected by digital electronic
circuitry in modern cameras (Fig. 18-15).
The third major cause of nonuniformity is local variations in the efficiency of the
camera in absorbing x- or gamma rays, such as manufacturing defects or damage
Digital
position
and
energy
signals
Y
Z correction
lookup table
Z
multiplied by
by correction f
f
Z' = f x Z
Z'
Corrected
energy
signal
to energy
discrimination
circuits
Z
FIGURE 18-15 Energy (Z) signal correction circuitry of a scintillation camera. The digital X- and Y-position
signals are used to look up a position-dependent correction factor. The uncorrected Z value is multiplied by
the correction factor. Finally, the corrected Z value is transmitted to the energy discrimination circuits.
694
to the collimator. These may be corrected by acquiring an image of an extremely

uniform planar source using a computer interfaced to the camera. A correction factor
is determined for each pixel in the image by dividing the average pixel count by that
pixel count. Each pixel in a clinical image is multiplied by its correction factor to
compensate for this cause of nonuniformity.
The digital spatial linearity and energy correction circuits of a modern scintillation camera do not directly affect its intrinsic spatial resolution. However, the obsolete
fully analog cameras, which lacked these corrections, used means such as thick light
pipes and light absorbers between the crystal and PMTs to improve spatial linearity
and uniformity. These reduced the amount of light reaching the PMTs. As mentioned
previously, the intrinsic spatial resolution and the energy resolution are largely determined by the statistics of the detection of light photons from the crystal. The loss of
signal from light photons in fully analog cameras significantly limited their intrinsic
spatial and energy resolutions. The use of digital linearity and energy correction permits camera designs that maximize light collection, thereby providing much better
intrinsic spatial resolution and energy resolution than fully analog cameras.
NEMA Specifications for Performance Measurements of

Scintillation Cameras
The National Electrical Manufacturers Association (NEMA) publishes a document
that specifies standard methods for measuring camera performance.3 All manufacturers of scintillation cameras publish NEMA performance measurements of their cameras, which are used by prospective purchasers in comparing cameras and for writing
purchase specifications. Before the advent of NEMA standards, manufacturers measured the performance of scintillation cameras in a variety of ways, making it difficult
to compare the manufacturers specifications for different cameras objectively. The
measurement methods specified by NEMA require specialized equipment and are not
intended to be performed by the nuclear medicine department. However, it is possible to perform simplified versions of the NEMA tests to determine whether a newly
purchased camera meets the published specifications. Unfortunately, the NEMA protocols omit testing of a number of important camera performance parameters.
Effects of Scatter, Collimator Spatial Resolution, and

Attenuation on Projection Images
Ideally, a nuclear medicine projection image would be a two-dimensional projection
of the three-dimensional activity distribution in the patient. If this were the case, the
number of counts in each point in the image would be proportional to the average
activity concentration along a straight line through the corresponding anatomy of
the patient. There are three main reasons why nuclear medicine images are not ideal
projection imagesattenuation of photons in the patient, inclusion of Compton
scattered photons in the image, and the degradation of spatial resolution with distance from the collimator. Furthermore, because most nuclear medicine images are
acquired over periods of many seconds or even minutes, patient motion, particularly
respiratory motion, is a source of image blurring.
Attenuation in the patient by Compton scattering and the photoelectric effect
prevents some photons that would otherwise pass through the collimator holes
from contributing to the image. The amount of attenuation is mainly determined
Performance Measures of Gamma Cameras, NEMA NU 1-2007.
695
by the path length through tissue and the densities of the tissues between a location in the patient and the corresponding point on the camera face. Thus, photons
from structures deeper in the patient are much more heavily attenuated than photons from structures closer to the camera face. Attenuation is more severe for lower
energy photons, such as the 68- to 80-keV characteristic x-rays emitted by Tl-201
( 0.19 cm1), than for higher energy photons, such as the 140-keV gamma rays
from Tc-99m ( 0.15 cm1) Nonuniform attenuation, especially in thoracic and
cardiac nuclear imaging, presents a particular problem in image interpretation.
The vast majority of the interactions with soft tissue by x- and gamma rays of the
energies used for nuclear medicine imaging are by Compton scattering. Some photons
that have scattered in the patient pass through the collimator holes and are detected. As
in x-ray transmission imaging such as radiography and CT, the relative number of scattered photons is greater when imaging thicker parts of the patient, such as the abdomen,
and the main effects of counts in the image from scattered photons are a loss of contrast
and an increase in random noise. As was mentioned earlier in this chapter, the number
of scattered photons contributing to the image is reduced by pulse height discrimination.
However, setting an energy window of sufficient width to encompass most of the photopeak permits a considerable amount of the scatter to contribute to image formation.
Operation and Routine Quality Control

Peaking a scintillation camera means to adjust its energy discrimination windows
to center them on the photopeak or photopeaks of the desired radionuclide. Modern cameras display the spectrum as a histogram of the number of interactions as a
function of pulse height (Fig. 18-16), like the display produced by a multichannel
analyzer (see Chapter 17). On this display, the energy window limits are shown as
vertical lines. A narrower energy window provides better scatter rejection, but also
reduces the number of unscattered photons recorded in the image.
Peaking may be performed manually by adjusting the energy window settings
while viewing the spectrum or automatically by the camera. Older scintillation cameras with mostly analog electronics required adjustment of the energy windows before
the first use each day and again before imaging another radionuclide. In modern scintillation cameras, the energy calibration and energy window settings are very stable.
For such cameras, peaking is typically adjusted and assessed only at the beginning
of each day of use with a single radionuclide, commonly using the Tc-99m or Co-57
source that is used for uniformity assessment. If other radionuclides are imaged during the day, preset energy windows for these radionuclides are used without adjusting
them for the actual energy spectra. Adjustment of the energy discrimination windows
for the photopeaks of other radionuclides (e.g., In-111, I-123, I-131, Xe-133, and
Tl-201) that might be imaged is performed infrequently, such as after major calibrations of the camera. A nearly scatter-free source should be used to assess or adjust the
energy windows; using the radiation emitted by a patient to assess or adjust the energy
windows would constitute poor practice because of its large scatter component.
The uniformity of the camera should be assessed daily and after each repair. The
assessment may be made intrinsically by using a Tc-99m point source, or the system uniformity may be evaluated by using a Co-57 planar source or a fillable flood
source. Uniformity images must contain sufficient counts so that quantum mottle
does not mask uniformity defects. The number of counts needed increases with the
useful area of the camera. As many as 5 to 15 million counts should be obtained for
the daily test of a modern large area, rectangular head camera. Uniformity images
can be evaluated visually or can be analyzed by a computer program, avoiding the
696
FIGURE 18-16 Displays of a pulse-height

spectrum used for peaking the scintillation camera (centering the pulse height
analyzer window on the photopeak). The
radionuclide is 99mTc. The energy window
is shown by the two vertical lines around
the photopeak. A. A properly adjusted
20% window. B. An improperly adjusted
20% window. C. A properly adjusted 15%
window.
s ubjectivity of visual evaluation. The uniformity test will reveal most malfunctions of
a scintillation camera. Figure 18-17 shows uniformity images from a modern scintillation camera. Multi-hole collimators are easily damaged by careless handling, such
as by striking them with an imaging table or dropping heavy objects on them. Technologists should inspect the collimators on each camera daily and whenever changing collimators. Old damage should be marked. The uniformity of each collimator
should be evaluated periodically, by using a Co-57 planar source or a fillable flood
source and whenever damage is suspected. The frequency of this testing depends on
the care taken by the technologists to avoid damaging the collimators and their reliability in visually inspecting the collimators and reporting new damage.
The spatial resolution and spatial linearity are typically assessed at least weekly. If
a four-quadrant bar phantom (Fig. 18-7) is used, it should be imaged four times with
a 90-degree rotation between images to ensure all quadrants of the camera are evaluated. If a parallel-line phantom is used, it should be imaged twice with a 90-degree
rotation between images.
The efficiency of each camera head should be measured periodically. It can be
monitored during uniformity tests by always performing the test with the source at
697
FIGURE 18-17 Uniformity images. A. Image from a modern high-performance camera with digital spatial linearity and energy signal correction circuitry. B. The other head of the same camera, with an optically
decoupled photomultiplier tube. (Images courtesy Carol Moll, CNMT)
the same distance, determining the activity of the source, recording the number of
counts and counting time, and calculating the count rate per unit activity.
Each camera should also have a complete evaluation at least annually, which
includes not only the items listed above, but also multienergy spatial registration and
count-rate performance. Table 19-1 in Chapter 19 contains a recommended schedule for quality control testing of scintillation cameras. Relevant tests should also be
performed after each repair or adjustment of a scintillation camera.
New cameras should receive acceptance testing by an experienced medical physicist to determine if the cameras performance meets the purchase specifications and
to establish baseline values for routine quality control testing.
Computers, Whole Body Scanning, and SPECT

Almost all scintillation cameras are connected to digital computers. The computer is
used for the acquisition, processing, and display of digital images and for control of
the movement of the camera heads. Computers used with scintillation cameras are
discussed later in this chapter.
Many large-FOV scintillation cameras can perform whole-body scanning. Some
systems move the camera past a stationary patient, whereas others move the patient
table past a stationary scintillation camera. Moving camera systems require less floor
space. In either case, the system must sense the position of the camera relative to the
patient table and must add values that specify the position of the camera relative to
the table to the X- and Y-position signals. Older systems with round or hexagonal
crystals usually had to make two passes over each side of the patient. Modern largearea rectangular head cameras can scan each side of all but extremely obese patients
with a single pass, saving considerable time and producing superior image statistics.
Many modern computer-equipped scintillation cameras have heads that can
rotate automatically around the patient and acquire images from different views.
The computer mathematically manipulates these images to produce cross-sectional
images of the activity distribution in the patient. This is called single photon emission computed tomography (SPECT) and is discussed in detail in Chapter 19.
698
Obtaining High-Quality Images

Attention must be paid to many technical factors to obtain high-quality images.
Imaging procedures should be optimized for each type of study. Sufficient counts
must be acquired so that quantum mottle in the image does not mask lesions. Imaging times must be as long as possible consistent with patient throughput and lack of
patient motion. The camera or table scan speed should be sufficiently slow during
whole-body scans to obtain adequate image statistics. The use of a higher resolution
collimator may improve spatial resolution and the use of a narrower energy window
to reduce scatter may improve image contrast.
Because the spatial resolution of a scintillation camera is degraded significantly
as the collimator-to-patient distance is increased, the camera heads should always
be as close to the patient as possible. In particular, thick pillows and mattresses on
imaging tables should be discouraged when images are acquired with the camera
head below the table.
Significant nonuniformities can be caused by careless treatment of the collimators. Collimators are easily damaged by collisions with imaging tables or by placing
them on top of other objects when they are removed from the camera.
Patient motion and metal objects worn by patients or in the patients clothing are
common sources of artifacts. Efforts should be made to identify metal objects and
to remove them from the patient or from the field of view. The technologist should
remain in the room during imaging to minimize patient motion. Furthermore, for
safety, a patient should never be left unattended under a moving camera, either in
whole-body or SPECT mode.
18.2 Computers in Nuclear Imaging

Most nuclear medicine computer systems consist of commercially available computers with additional components to enable them to acquire, process, and display
images from scintillation cameras. A manufacturer may incorporate a computer for
image acquisition and camera control in the camera itself and provide a separate
computer for image processing and display, or provide a single computer for both
purposes.
Modern scintillation cameras create the pairs of X- and Y-position signals using
digital circuitry and transfer them to the computer in digital form. The formation of
a digital image is described in the following section.
Digital Image Formats in Scintillation Camera Imaging

As discussed in Chapter 5, a digital image consists of a rectangular array of numbers,
and an element of the image represented by a single number is called a pixel. In
nuclear medicine, each pixel represents the number of counts detected from activity in a specific location within the patient. Common image formats are 642 and
1282 pixels, reflecting the low spatial resolution of scintillation cameras. Whole-body
images are stored in larger formats, such as 256 by 1,024 pixels. If one byte (8 bits) is
used for each pixel, the image is said to be in byte mode; if two bytes are used for each
pixel, the image is said to be in word mode. A single pixel can store as many as 255
counts in byte mode and 65,535 counts in word mode. Modern nuclear medicine
computers may allow only word-mode images.
699
Image Acquisition
Frame-Mode Acquisition
Image data in nuclear medicine are acquired in either frame or list mode. Figure18-18
illustrates frame-mode acquisition. Before acquisition begins, a portion of the computers memory is designated to contain the image. All pixels within this image are
set to zero. After acquisition begins, pairs of X- and Y-position signals are received
from the camera, each pair corresponding to the detection of a single x- or gamma
ray. Each pair of numbers designates a single pixel in the image. One count is added
to the counts in that pixel. As many pairs of position signals are received, the image
is formed.
There are three types of frame-mode acquisition: static, dynamic, and gated. In
a static acquisition, a single image is acquired for either a preset time interval or
until the total number of counts in the image reaches a preset number. In a dynamic
acquisition, a series of images is acquired one after another, for a preset time per
image. Dynamic image acquisition is used to study dynamic processes, such as the
first transit of a bolus of a radiopharmaceutical through the heart or the extraction
and excretion of a radiopharmaceutical by the kidneys.
Some dynamic processes occur too rapidly for them to be effectively portrayed
by dynamic image acquisition; each image in the sequence would have too few
counts to be statistically valid. However, if the dynamic process is repetitive, gated
image acquisition may permit the acquisition of an image sequence that accurately
Pairs of digital
position signals
X
All pixels of
image are
set to zero.
Image memory
3
First
interaction
2
Second
interaction
1
Third
interaction
2 0
1 0
Video interface
Video monitor
DAC
DAC
DAC
DAC
FIGURE 18-18 Acquisition of an image in frame mode. This figure illustrates the incorporation of position data
from the first three interactions into an image. This example shows a 4 by 4 pixel image for the purpose of illustration. Actual nuclear medicine projection images are typically acquired in formats of 64 x 64 or 128 x 128 pixels.
700
depicts the dynamic process. Gated acquisition is most commonly used for evaluating cardiac mechanical performance in cardiac blood pool studies and, sometimes,
in myocardial perfusion studies. Gated acquisition requires a physiologic monitor
that provides a trigger pulse to the computer at the beginning of each cycle of the
process being studied. In gated cardiac studies, an electrocardiogram (ECG) monitor provides a trigger pulse to the computer whenever the monitor detects a QRS
complex.
Figure 18-19 depicts the acquisition of a gated cardiac image sequence. First,
space for the desired number of images (usually 16 to 24 for gated cardiac blood
pool imaging) is reserved in the computers memory. Next, several cardiac cycles are
timed and the average time per cycle is determined. The time per cycle is divided
by the number of images in the sequence to obtain the time T per image. Then the
acquisition begins. When the first trigger pulse is received, the acquisition interface
sends the data to the first image in the sequence for a time T. Then it is stored in
the second image in the sequence for a time T, after which it is stored in the third
image for a time T. This process proceeds until the next trigger pulse is received.
Then the process begins anew, with the data being added to that in the first image
for a time T, then the second image for a time T, etc. This process is continued until
a preset time interval, typically 10 min, has elapsed, enabling sufficient counts to be
collected for the image sequence to form a statistically valid depiction of an average
cardiac cycle.
List-Mode Acquisition
In list-mode acquisition, the pairs of X- and Y-position values are stored in a list
(Fig. 18-20), instead of being immediately formed into an image. Periodic timing
marks are included in the list. If a physiologic monitor is being used, as in gated
cardiac imaging, trigger marks are also included in the list. After acquisition is complete, the list-mode data are reformatted into conventional images for display. The
advantage of list-mode acquisition is that it allows great flexibility in how the X and
Y values are combined to form images. The disadvantages of list-mode acquisition
are that it generates large amounts of data, requiring more memory to acquire and
QRS
QRS
First
image
Second
image
QRS
Third
image
Fourth
image
FIGURE 18-19 Acquisition of a gated cardiac image sequence. Only four images are shown here. Sixteen
to twenty-four images are typically acquired.

X
Y
701
Image acquisition memory
ADC
X1
Y1
X2
Y2
X3
Y3
X7
Y7
X8
Y8 time X9
X4
Y4 time X5
Y5
X6
Y6
ADC
Y9 X10 Y10 X11 Y11 X12 Y12
X13 Y13 X14 Y14 time X15 Y15
...
FIGURE 18-20 Acquisition of image data in list mode. The digital X- and Y-position signals are stored as a
list. Periodically, a timing signal is also stored. The image or images cannot be viewed until the list-mode data
are formed into an image matrix or matrices.
disk space to store than frame-mode images, and that the data must be subsequently
processed into standard images for viewing.
Image Processing in Nuclear Medicine

A major reason for the use of computers in nuclear medicine is that they provide the
ability to present the data in the unprocessed images in ways that are of greater use to the
clinician. Although it is not within the scope of this section to provide a comprehensive
survey of image processing in nuclear medicine, the following are common examples.
Image Subtraction
When one image is subtracted from another, each pixel count in one image is subtracted from the corresponding pixel count in the other. Negative numbers resulting
from these subtractions are set to zero. The resultant image depicts the change in
activity that occurs in the patient during the time interval between the acquisitions
of the two images.
Regions of Interest and Time-Activity Curves

A region of interest (ROI) is a closed boundary that is superimposed on the image. It
may be drawn manually or it may be drawn automatically by the computer. The sum
of the counts in all pixels in the ROI is an indication of the activity in the corresponding portion of the patient.
To create a time-activity curve (TAC), a ROI must first be drawn on one image of
a dynamic or gated image sequence. The same ROI is then superimposed on each
image in the sequence by the computer and the total number of counts within the
ROI is determined for each image. Finally, the counts within the ROI are plotted as a
function of image number. The resultant curve is an indication of the activity in the
corresponding portion of the patient as a function of time. Figure 18-21 shows ROIs
over both kidneys of a patient and TACs describing the uptake and excretion of the
radiopharmaceutical Tc-99m MAG-3 by the kidneys.
Spatial Filtering
Nuclear medicine images have a grainy appearance because of the statistical nature of
the acquisition process. This quantum mottle can be reduced by a type of spatial filtering called smoothing. Unfortunately, smoothing also reduces the spatial resolution
of the image. Images should not be smoothed to the extent that clinically significant
detail is lost. Smoothing is a type of convolution filtering, which is described in detail
in Chapter 4 and Appendix G.
702
FIGURE 18-21 Regions of interest (bottom) and time-activity curves (top).
Left Ventricular Ejection Fraction

The left ventricular ejection fraction (LVEF) is a measure of the mechanical performance of the left ventricle of the heart. It is defined as the fraction of the end-diastolic
volume ejected during a cardiac cycle:

LVEF (VED VES ) / VED
[18-10]
where VED is the end-diastolic volume and VES is the end-systolic volume of the ventricle. In nuclear medicine, it is can be determined from an equilibrium-gated blood
pool image sequence, using Tc-99mlabeled red blood cells. The image sequence is
acquired from a left anterior oblique (LAO) projection, with the camera positioned
at the angle demonstrating the best separation of the two ventricles, after sufficient
time has elapsed for the administered activity to reach a uniform concentration in
the blood.
The calculation of the LVEF is based on the assumption that the counts from
left ventricular activity are approximately proportional to the ventricular volume
throughout the cardiac cycle. A ROI is first drawn around the left ventricular
cavity, and a TAC is obtained by superimposing this ROI over all images in the
sequence. The first image in the sequence depicts end diastole, and the image containing the least counts in the ROI depicts end systole. The total left ventricular
703
FIGURE 18-22 End-diastolic (top) and end-systolic (bottom) images from a gated cardiac blood-pool
study, showing ROIs used to determine the left ventricular ejection fraction (LVEF). The small dark ROI below
and to the right of the LV ROI in the end-systolic image is used to determine and correct for the extracardiac
(background) counts.
counts in the end-diastolic and end-systolic images are determined. Some programs use the same ROI around the left ventricle for both images, whereas, in
other programs, the ROI is drawn separately in each image to better fit the varying shape of the ventricle. Unfortunately, each of these counts is due not only to
activity in the left ventricular cavity but also to activity in surrounding tissues,
chambers, and great vessels. To compensate for this crosstalk (commonly called
background activity), another ROI is drawn just beyond the wall of the left
ventricle, avoiding active structures such as the spleen, cardiac chambers, and
great vessels. The number of counts in the left ventricular ROI due to crosstalk is
estimated as follows:

Counts crosstalk
(Counts in crosstalk ROI) (Pixels in LV ROI)

[18-11]
(Pixels in crosstalk ROI)
Figure 18-22 shows end-diastolic and end-systolic images of the heart from the LAO
projection and the ROIs used to determine the LVEF. The LVEF is then calculated
using the following equation:

LVEF
(Counts ED) (Counts ES)

(Counts ED) (Counts crosstalk)
[18-12]
704
Summary
This chapter described the principles of operation of the Anger scintillation camera, which forms images, depicting the distribution of x- and gamma-rayemitting
radionuclides in patients, using a planar crystal of the scintillator NaI(Tl) optically
coupled to a two-dimensional array of photomultiplier tubes. The collimator, necessary for formation of projection images, imposes a compromise between spatial resolution and sensitivity in detecting x- and gamma rays. Because of the collimator, the
spatial resolution of the images degrades with distance from the face of the camera.
In Chapter 19, the use of the scintillation camera to perform computed tomography,
called SPECT, is described.
SUGGESTED READINGS
Anger HO. Radioisotope cameras. In: Hine GJ, ed. Instrumentation in nuclear medicine, vol. 1. New York,
NY: Academic Press, 1967:485552.
Gelfand MJ, Thomas SR. Effective use of computers in nuclear medicine. New York, NY: McGraw-Hill, 1988.
Graham LS, Levin CS, Muehllehner G. Anger scintillation camera. In: Sandler MP, et al., eds. Diagnostic
nuclear medicine. 4th ed. Baltimore, MD: Lippincott Williams & Wilkins, 2003:3142.
Groch MW. Cardiac function: gated cardiac blood pool and first pass imaging. In: Henkin RE, et al., eds.
Nuclear medicine, vol. 1. St. Louis, MO: Mosby, 1996:626643.
Simmons GH, ed. The scintillation camera. New York, NY: Society of Nuclear Medicine, 1988.
Yester MV, Graham LS, eds. Advances in nuclear medicine: the medical physicists perspective. Proceedings of
the 1998 Nuclear Medicine Mini Summer School, American Association of Physicists in Medicine,
June 2123,1998, Madison, WI.
Chapter
19
Nuclear ImagingEmission
Tomography
The formation of projection images in nuclear medicine was discussed in the previous chapter. A nuclear medicine projection image depicts a two-dimensional projection of the three-dimensional activity distribution in the patient. The disadvantage of
a projection image is that the contributions to the image from structures at different
depths overlap, hindering the ability to discern the image of a structure at a particular depth. Tomographic imaging is fundamentally differentit attempts to depict the
activity distribution in a single cross section of the patient.
There are two fundamentally different types of tomography: conventional tomography, also called geometric or focal plane tomography, and computed tomography.
In conventional tomography, structures out of a focal plane are not removed from
the resultant image; instead, they are blurred by an amount proportional to their
distances from the focal plane. Those close to the focal plane suffer little blurring
and remain apparent in the image. Even those farther away, although significantly
blurred, contribute to the image, thereby reducing contrast and adding noise. In
distinction, computed tomography uses mathematical methods to remove overlying structures completely. Computed tomography requires the acquisition of a set
of projection images from at least a 180-degree arc about the patient. The projection
image information is then mathematically processed by a computer to form images
depicting cross sections of the patient. Just as in x-ray transmission imaging, both
conventional and computed tomography are possible in nuclear medicine imaging.
Both single photon emission computed tomography (SPECT) and positron emission
tomography (PET) are forms of computed tomography.
19.1 Focal Plane Tomography in Nuclear Medicine

Focal plane tomography once had a significant role in nuclear medicine, but is
seldom used today. The rectilinear scanner, when used with focused collimators,
is an example of conventional tomography. A number of other devices have been
developed to exploit conventional tomography in nuclear medicine. The Anger
tomoscanner used two small scintillation cameras with converging collimators, one
above and one below the patient table, to scan the patient in a raster pattern; a
single scan produced multiple whole-body images, each showing structures at a
different depth in the patient in focus. The seven-pinhole collimator was used with
a conventional scintillation camera and computer to produce short-axis images
of the heart, each showing structures at a different depth in focus. The rectilinear
scanner and the Anger tomoscanner are no longer produced. The seven-pinhole
collimator, which never enjoyed wide acceptance, has been almost entirely displaced by SPECT.
705
706
The scintillation camera itself, when used for planar imaging with a parallelhole collimator, produces a weak tomographic effect. The system spatial resolution
decreases with distance, causing structures farther from the camera to be more
blurred than closer structures. Furthermore, attenuation of photons increases
with depth in the patient, also enhancing the visibility of structures closer to the
camera. This effect is perhaps most clearly evident in planar skeletal imaging of
the body. In the anterior images, for example, the sternum and anterior portions
of the ribs are clearly shown, whereas the spine and posterior ribs are barely
evident.
19.2 Single Photon Emission Computed Tomography

Single photon emission computed tomography (SPECT) generates transverse
images depicting the distribution of x- or gamma-rayemitting nuclides in
patients. Standard planar projection images are acquired from an arc of 180
degrees (most cardiac SPECT) or 360 degrees (most noncardiac SPECT) about
the patient. Although these images could be obtained by any collimated imaging
device, the vast majority of SPECT systems use one or more scintillation camera heads that revolve about the patient. The SPECT systems digital computer
then reconstructs the transverse images using either filtered backprojection or
an iterative reconstruction method, which are described later in this chapter, as
does the computer in an x-ray CT system. Figure19-1 shows a variety of SPECT
systems.
SPECT was invented by David Kuhl and others in the early 1960s, about 10 years
before the invention of x-ray CT by Hounsfield (Kuhl and Edwards, 1963). However,
in contrast to x-ray CT, most features of interest in SPECT images were also visible
in planar nuclear medicine images and SPECT did not come into routine clinical use
until the late 1980s.
Image Acquisition
The camera head or heads of a SPECT system revolve about the patient, acquiring projection images. The head or heads may acquire the images while moving
(continuous acquisition) or may stop at predefined evenly spaced angles to acquire
the images (step and shoot acquisition). If the camera heads of a SPECT system
produced ideal projection images (i.e., no attenuation by the patient and no degradation of spatial resolution with distance from the camera), projection images from
opposite sides of the patient would be mirror images and projection images over a
180-degree arc would be sufficient for transverse image reconstruction. However,
in SPECT, attenuation greatly reduces the number of photons from activity in the
half of the patient opposite the camera head and this information is greatly blurred
by the distance from the collimator. Therefore, for most noncardiac studies, such as
bone SPECT, the projection images are acquired over a complete revolution (360
degrees) about the patient. However, most nuclear medicine laboratories acquire
cardiac SPECT studies, such as myocardial perfusion studies, over a 180-degree arc
symmetric about the heart, typically from the 45-degree right anterior oblique view
to the 45-degree left posterior oblique view (Fig. 19-2). The 180-degree acquisition produces reconstructed images of superior contrast and resolution, because
Chapter 19 Nuclear ImagingEmission Tomography
707
the projection images of the heart from the opposite 180 degrees have poor spatial
resolution and contrast due to greater distance and attenuation. Although studies
have shown that the 180-degree acquisition can introduce artifacts (Liu et al., 2002),
the 180-degree acquisition is more commonly used than the 360-degree acquisition
for cardiac studies.
SPECT projection images are usually acquired in either a 642 or a 1282 pixel format. Using too small a pixel format reduces the spatial resolution of the projection
images and of the resultant reconstructed transverse images. When the 642 format is
used, typically 60 or 64 projection images are acquired and, when a 1282 format is
chosen, 120 or 128 projection images are acquired. Using too few projections creates
radial streak artifacts in the reconstructed transverse images.
The camera heads on older SPECT systems followed circular orbits around the
patient while acquiring images. Circular orbits are satisfactory for SPECT imaging of
the brain, but cause a loss of spatial resolution in body imaging because the circular
FIGURE 19-1 A. SPECT/CT system with two scintillation camera heads in a fixed 180-degree orientation
and a non-diagnostic x-ray CT system for attenuation correction and anatomic correlation. The x-ray source
is on the right side of the gantry and a flat-panel x-ray image receptor is on the left. B. Technologist moving
the upper camera head closer to the patient for SPECT imaging. Photo credit: Emi Manning UC Davis Health
System
708
FIGURE 19-1 (Continued) C. Dual head, variable angle SPECT/CT camera with heads in the 90-degree
orientation for cardiac SPECT and in the 180-degree orientation (D) for other SPECT or whole body planar
imaging. E. Dual head, fixed 90-degree SPECT camera for cardiac imaging (Image courtesy Siemens Medical
Solutions). F. Single head SPECT camera, with head in a position for planar imaging (Image courtesy Emi
Manning UC Davis Health System).
FIGURE 19-2 180-degree cardiac orbit.
RAO
180
LPO
Circular orbit
709
Body contouring orbit
FIGURE 19-3 Circular (A) and body-contouring (B) orbits.
orbit causes the camera head to be many centimeters away from the surface of the
body during the anterior and perhaps the posterior portions of its orbit (Fig. 19-3).
Modern SPECT systems provide noncircular orbits (also called body contouring)
that keep the camera heads in close proximity to the surface of the body throughout
the orbit. For some systems, the technologist specifies the noncircular orbit by placing the camera head as close as possible to the patient at several angles, from which
the cameras computer determines the orbit. Other systems perform automatic body
contouring, using sensors on the camera heads to determine their proximity to the
patient at each angle.
In brain SPECT, it is usually possible for the camera head to orbit with a much
smaller radius than in body SPECT, thereby producing images of much higher spatial resolution. In many older cameras, a large distance from the physical edge of
the camera head to the useful portion of the detector often made it impossible to
orbit at a radius within the patients shoulders while including the base of the brain
in the images. These older systems were therefore forced to image the brain with an
orbit outside the patients shoulders, causing a significant loss of resolution. Most
modern SPECT systems permit brain imaging with orbits within the patients shoulders, although a patients head holder extending beyond the patient table is generally
necessary.
Transverse Image Reconstruction

After the projection images are acquired, they are usually corrected for nonuniformities and for center-of-rotation (COR) misalignments. (These corrections are
discussed below, Quality Control in SPECT.) Following these corrections, transverse image reconstruction is performed using either filtered backprojection or
iterative methods.
As described in Chapter 10, filtered backprojection consists of two steps. First,
the projection images are mathematically filtered. Then, to form a particular transverse image, simple backprojection is performed of the row, corresponding to that
transverse image, of each projection image. For example, the fifth row of each projection image is backprojected to form the fifth transverse image. A SPECT study
produces transverse images covering the entire field of view of the camera in the axial
direction from each revolution of the camera head or heads.
FIGURE 19-4 Typical filter kernels used

for filtered backprojection. The kernels are
shown in frequency space. Filter Kernel A
is a Butterworth filter of fifth order with a
critical frequency of 0.15 Nyquist and Filter
Kernel B is a Butterworth filter of fifth order
with a critical frequency of 0.27 Nyquist. Filter Kernel A provides more smoothing than
Filter Kernel B. A ramp filter, which provides
no smoothing, is also shown.
0.5
0.4
Amplitude
710
ramp
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
Spatial frequency (cycles/pixel)
0.5
Mathematical theory specifies that the ideal filter kernel, when displayed in
the spatial frequency domain, is the ramp filter (Fig. 19-4). (The spatial frequency
domain is discussed in Appendix G, Convolution and Fourier Transforms.) However,
the actual projection images contain considerable statistical noise, from the random
nature of radioactive decay and photon interactions, due to the relatively small number of counts in each pixel. If the images were filtered using a ramp filter kernel and
then backprojected, the resultant transverse images would contain an unacceptable
amount of statistical noise.
In the spatial frequency domain, statistical noise predominates in the high-frequency portion. Furthermore, the spatial resolution characteristics of the scintillation
camera cause a reduction of higher spatial frequency information that increases with
distance of the structure being imaged from the camera. To smooth the projection
images before backprojection, the ramp filter kernel is modified to roll off at higher
spatial frequencies. Unfortunately, this reduces the spatial resolution of the projection images and thus of the reconstructed transverse images. A compromise must
therefore be made between spatial resolution and statistical noise of the transverse
images.
Typically, a different filter kernel is selected for each type of SPECT study; for
example, a different kernel would be used for HMPAO brain SPECT than would be
used for Tl-201 myocardial perfusion SPECT. The choice of filter kernel for a particular type of study is determined by the amount of statistical noise in the projection
images (mainly determined by the injected activity, collimator, and acquisition time
per image) and their spatial resolution (determined by the collimator and distances
of the camera head(s) from the organ being imaged). The preference of the interpreting physician regarding the appearance of the images also plays a role. Projection
images of better spatial resolution and less quantum mottle require a filter with a
higher spatial frequency cutoff to avoid unnecessary loss of spatial resolution in the
reconstructed transverse images, whereas projection images of poorer spatial resolution and greater quantum mottle require a filter with a lower spatial frequency cutoff
to avoid excessive quantum mottle in the reconstructed transverse images. Although
the SPECT cameras manufacturer may suggest filters for specific imaging procedures,
the filters are usually empirically optimized in each nuclear medicine laboratory. Figure 19-5 shows a SPECT image created using three different filter kernels, illustrating
too much smoothing, proper smoothing, and no smoothing.
Filtered backprojection is computationally efficient. However, it is based upon the
assumption that the projection images are perfect projections of a three-dimensional
711
FIGURE 19-5 SPECT images created by filtered backprojection. The projection images were filtered using
the filter kernels shown in Figure 19-4. The image on the left, produced using Filter Kernel A, exhibits a significant loss of spatial resolution. The image in the center was produced using Filter Kernel B, which provides
a proper amount of smoothing. The image on the right, produced using the ramp filter, shows good spatial
resolution, but excessive statistical noise.
object. As discussed in the previous chapter, this is far from true in scintillation camera imaging, mainly because of attenuation of photons in the patient, inclusion of
Compton scattered photons in the image, and the degradation of spatial resolution
with distance from the collimator.
In SPECT, iterative reconstruction methods are increasingly being used instead
of filtered backprojection. In iterative methods, an initial activity distribution in the
patient is assumed. Then, projection images are calculated from the assumed activity
distribution, using a model of the imaging characteristics of the scintillation camera
and the patient. The calculated projection images are compared with the actual projection images and, based upon this comparison, the assumed activity distribution is
adjusted. This process is repeated several times, with successive adjustments to the
assumed activity distribution, until the calculated projection images approximate the
actual projection images (Fig. 19-6).
As was stated above, in each iteration, projection images are calculated from the
assumed activity distribution. The calculation of projection images can use the point
spread function of the scintillation camera, which takes into account the decreasing
spatial resolution with distance from the camera face. The point spread function can be
modified to incorporate the effect of photon scattering in the patient. Furthermore, if a
map of the attenuation characteristics of the patient is available, the calculation of the
projection images can include the effects of attenuation. If this is done, iterative methods
will partially compensate for the effects of decreasing spatial resolution with distance,
photon scattering in the patient, and attenuation in the patient. Iterative reconstruction
can be used to produce higher quality tomographic images than filtered backprojection, or it can be used to produce images of similar quality to those produced by filtered
backprojection, but with less administered activity or shorter acquisition times.
Iterative methods are computationally less efficient than filtered backprojection.
However, the increasing speed of computers, the small image matrix sizes used in
nuclear imaging, and development of computationally efficient algorithms, such as
the ordered-subset expectation maximization method (Hudson and Larkin, 1994),
have made iterative reconstruction feasible for SPECT. Three-dimensional spatial filtering is commonly performed after iterative reconstruction to reduce image noise.
712
FIGURE 19-6 Flowchart for iterative reconstruction. In some implementations, iterative reconstruction
is performed for a specified number
of iterations, instead of being terminated when a sufficiently good
approximation is achieved.
Assume Activity
Distribution
Calculate Projection
Images
Compare to Actual
Projection Images
Adequate
Match Between
Calculated and Actual
Projection
Images?
Yes
No
Adjust Assumed
Activity Distribution
Stop
Attenuation Correction in SPECT

Radioactivity whose x- or gamma rays must traverse long paths through the patient
produces fewer counts, due to attenuation, than does activity closer to the surface
of the patient adjacent to the camera. For this reason, transverse image slices of a
phantom with a uniform activity distribution, such as a cylinder filled with a wellmixed solution of radionuclide, will show a gradual decrease in activity toward the
center (Fig. 19-7, on the left). Attenuation effects are more severe in body SPECT
than in brain SPECT.
FIGURE 19-7 Attenuation correction. On the left is a reconstructed transverse image slice of a cylindrical
phantom containing a well-mixed radionuclide solution. This image shows a decrease in activity toward the
center due to attenuation. (A small ring artifact, unrelated to the attenuation, is also visible in the center of the
image.) In the center is the same image corrected by the Chang method, using a linear attenuation coefficient
of 0.12 cm1, demonstrating proper attenuation correction. On the right is the same image, corrected by the
Chang method using an excessively large attenuation coefficient.
713
Approximate methods are available for attenuation correction. One of the most
common, the Chang method, presumes a constant attenuation coefficient throughout
the patient (Chang, 1978). Approximate attenuation corrections can overcompensate
or undercompensate for attenuation. If such a method is to be used, its proper functioning should be verified using phantoms before its use in clinical studies.
However, attenuation is not uniform in the patient, particularly in the thorax.
These approximate methods cannot compensate for nonuniform attenuation.
Several manufacturers provide SPECT cameras with sealed radioactive sources
(commonly containing Gd-153, which emits 97 and 103-keV gamma rays) to measure the attenuation through the patient. The sources are used to acquire transmission data from projections around the patient. After acquisition, the transmission
projection data are reconstructed to provide maps of tissue attenuation characteristics across transverse sections of the patient, similar to x-ray CT images. Finally, these
attenuation maps are used during an iterative SPECT image reconstruction process to
provide attenuation-corrected SPECT images.
The transmission sources are available in several configurations. These include
scanning collimated line sources that are used with parallel-hole collimators, arrays
of fixed line sources used with parallel-hole collimators, and a fixed line source
located at the focal point of a fan-beam collimator.
The transmission data are usually acquired simultaneously with the acquisition
of the emission projection data, because performing the two separately can pose significant problems in the spatial alignment of the two data sets and greatly increases
the total imaging time. The radionuclide used for the transmission measurements is
chosen to have primary gamma ray emissions that differ significantly in energy from
those of the radiopharmaceutical. Separate energy windows are used to differentiate
the photons emitted by the transmission source from those emitted by the radiopharmaceutical. However, scattering of the higher energy photons in the patient and
in the detector causes some cross-talk in the lower energy window.
Major manufacturers of nuclear medicine imaging systems provide systems combining two scintillation camera heads capable of planar imaging and SPECT and an
x-ray CT scanner, with a single patient bed. These are referred to as SPECT/CT systems. In SPECT/CT systems, the x-ray CT attenuation image data can be used to correct the radionuclide emission data for attenuation by the patient. This is discussed
in more detail later in this chapter.
Attenuation correction using radioactive transmission sources and x-ray CTderived attenuation maps has been extensively studied in myocardial perfusion
SPECT, where attenuation artifacts can mimic perfusion defects. These studies have
shown that attenuation correction reduces attenuation artifacts and produces modest
improvement in diagnostic performance when the studies are read by experienced
clinicians (Hendel et al., 2002; Masood et al., 2005). However, other studies have
shown that attenuation correction can cause artifacts, particularly when there is spatial misalignment of the emission data with respect to the attenuation maps determined from the transmission information. It remains common for SPECT myocardial
perfusion imaging to be performed without attenuation correction, although that
may change with the increasing implementation of SPECT/CT.
Generation of Coronal, Sagittal, and Oblique Images

The pixels from the transverse image slices may be reordered to produce coronal
and sagittal images. For cardiac imaging, it is desirable to produce oblique images
714
oriented either parallel (vertical and horizontal long-axis images) or perpendicular

(short-axis images) to the long axis of the left ventricle. Because there is considerable anatomic variation among patients regarding the orientation of the long axis
of the left ventricle, the long axis of the heart must be determined before the computer can create the oblique images. This task is commonly performed manually
by a technologist.
Collimators for SPECT

Most SPECT is performed using parallel-hole collimators. However, specialized
collimators have been developed for SPECT. The fan-beam collimator, shown in
Figure19-8, is a hybrid of the converging and parallel-hole collimator. Because it
is a parallel-hole collimator in the y-direction, each row of pixels in a projection
image corresponds to a single transaxial slice of the subject. In the x-direction, it is a
converging collimator, with spatial resolution and efficiency characteristics superior
to those of a parallel-hole collimator (see Fig. 18-11). Because a fan-beam collimator is a converging collimator in the cross-axial direction, its field-of-view (FOV)
decreases with distance from the collimator. For this reason, the fan-beam collimator
is mainly used for brain SPECT; if the collimator is used for body SPECT, portions
of the body are excluded from the FOV, which can cause artifacts, called truncation
artifacts, in the reconstructed images.
Multihead SPECT Cameras

To reduce the limitations imposed on SPECT by collimation and limited time per
view, camera manufacturers provide SPECT systems with two scintillation camera
heads that revolve about the patient (Fig. 19-1) and, in the past, SPECT systems
with three heads were commercially available from at least two manufacturers. The
use of multiple camera heads permits the use of higher resolution collimators, for a
given level of quantum mottle in the tomographic images, than would a single head
system. However, the use of multiple camera heads poses considerable technical
challenges for the manufacturer. It places severe requirements upon the electrical
FIGURE 19-8 Fan-beam collimator.
715
and mechanical stability of the camera heads. In particular, the X and Y offsets and
X and Y magnification factors of all the heads must be precisely matched throughout
the rotation about the patient. Todays multihead systems are very stable and provide
high-quality tomographic images for a variety of clinical applications.
Multihead scintillation cameras are available in several configurations. Double-head cameras with opposed heads (180 degrees head configuration) are good
for head and body SPECT and whole-body planar scans (Figure 19-1 A and B).
Triple-head, fixed-angle cameras are good for head and body SPECT, but less suitable
for whole-body planar scans because of the limited width of the crystals. Doublehead, variable-angle cameras are highly versatile, capable of head and body SPECT
and whole-body planar scans with the heads in the 180-degree configuration and
cardiac SPECT in the 90-degree configuration (Figure 19-1, C and D). (The useful
portion of the NaI crystal does not extend all the way to the edge of a camera head.
If the two camera heads are placed at an angle of exactly 90 degrees to each other,
both heads cannot be close to the patient without parts of the patient being outside
of the fields of view. For this reason, one manufacturer provides the option of SPECT
acquisitions with the heads at a 76-degree angle to each other.)
Performance
Spatial Resolution
The spatial resolution of a SPECT system can be measured by acquiring a SPECT study
of a line source, such as a capillary tube filled with a solution of Tc-99m, placed parallel
to the axis of rotation. The National Electrical Manufacturers Association (NEMA) has a
protocol for measuring spatial resolution in SPECT. This protocol specifies a cylindrical
plastic water-filled phantom, 22 cm in diameter, containing three line sources (Fig. 19-9,
on the left) for measuring spatial resolution. The full widths at half maximum (FWHMs)
of the line sources are measured from the reconstructed transverse images, as shown on
the right in Figure 19-9. A ramp filter is used in the filtered backprojection, so that the
filtering does not reduce the spatial resolution. The NEMA spatial resolution measurements are primarily determined by the collimator used. The tangential resolution for the
peripheral sources (typically 7 to 8 mm FWHM for low-energy high-resolution parallelhole collimators) is superior to both the central resolution (typically 9.5 to 12 mm) and
the radial resolution for the peripheral sources (typically 9.4 to 12 mm).
These FWHMs measured using the NEMA protocol, while providing a useful
index of ultimate system performance, are not necessarily representative of clinical
performance, because these spatial resolution studies can be acquired using longer
imaging times and closer orbits than would be possible in a patient. Patient studies
may require the use of lower resolution (higher efficiency) collimators than the one
used in the NEMA measurement to obtain adequate image statistics. In addition, the
filters used before backprojection for clinical studies cause more blurring than do
the ramp filters used in NEMA spatial resolution measurements. The NEMA spatial
resolution measurements fail to show the advantage of SPECT systems with two or
three camera heads; double and triple head cameras will permit the use of higher
resolution collimators for clinical studies than will single head cameras.
Spatial resolution deteriorates as the radius of the camera orbit increases. For this
reason, brain SPECT produces images of much higher spatial resolution than does
body SPECT. For optimal spatial resolution, the SPECT camera heads should orbit
the patient as closely as possible. Body-contouring orbits (see above, Design and
Operation) provide better resolution than circular orbits.
716

Radial
Tangential
Center
Radial
Co-57 line sources
Tangential
FIGURE 19-9 NEMA phantom for evaluating the spatial resolution of a SPECT camera (left). The phantom
is a 22-cm diameter plastic cylinder, filled with water and containing three Co-57 line sources. One line source
lies along the central axis and the other two are parallel to the central line source, 7.5 cm away. A SPECT study
is acquired with a camera radius of rotation (distance from collimator to AOR) of 15 cm. The spatial resolution
is measured from reconstructed transverse images, as shown on the right. Horizontal and vertical profiles are
taken through the line sources and the FWHMs of these LSFs are determined. The average central resolution
(FWHM of the central LSF) and the average tangential and radial resolutions (determined from the FWHMs of
the two peripheral sources as shown) are determined. (Adapted from Performance Measurements of Scintillation Cameras, National Electrical Manufacturers Association, 2001.)
Comparison of SPECT to Conventional Planar Scintillation

Camera Imaging
In theory, SPECT should produce spatial resolution similar to that of planar scintillation camera imaging. In clinical imaging, its resolution is usually slightly worse. The
camera head is usually closer to the patient in conventional planar imaging than in
SPECT. The spatial filtering used in SPECT to reduce statistical noise also reduces
spatial resolution. The short time per view of SPECT may mandate the use of a lower
resolution collimator to obtain adequate numbers of counts.
In planar nuclear imaging, radioactivity in tissues in front of and behind an organ
or tissue of interest causes a reduction in contrast. Furthermore, if the activity in
these overlapping structures is not uniform, the pattern of this activity distribution
is superimposed on the activity distribution in the organ or tissue of interest. As
such, it is a source of structural noise that impedes the ability to discern the activity
distribution in the organ or tissue of interest. The main advantage of SPECT over
conventional planar nuclear imaging is improved contrast and reduced structural
noise produced by eliminating counts from the activity in overlapping structures.
SPECT using iterative reconstruction can also partially compensate for the effects of
the scattering of photons in the patient and collimator effects such as the decreasing
spatial resolution with distance from the camera and collimator septal penetration.
When attenuation is measured using sealed radioactive transmission sources or an
x-ray CT scanner, SPECT can partially compensate for the effects of photon attenuation in the patient.
717
Quality Control in SPECT

Even though a technical quality control program is important in planar nuclear
imaging, it is critical to SPECT. Equipment malfunctions or maladjustments that
would not noticeably affect planar images can markedly degrade the spatial resolution of SPECT images and produce significant artifacts, some of which may mimic
pathology. Upon installation, a SPECT camera should be tested by a medical physicist. Following acceptance testing, a quality control program should be established to
ensure that the systems performance remains comparable to its initial performance.
X and Y Magnification Factors and Multienergy Spatial

Registration
The X and Y magnification factors, often called X and Y gains, relate distances in the
object being imaged, in the x and y directions, to the numbers of pixels between the
corresponding points in the resultant image. The X magnification factor is determined from an image of two point sources placed against the cameras collimator a
known distance apart along a line parallel to the x-axis:
X mag
actual distance between centers of point sources

number of pixels between centers of point sources
The Y magnification factor is determined similarly but with the sources parallel to
the y-axis. The X and Y magnification factors should be equal. If they are not, the
projection images will be distorted in shape, as will be coronal, sagittal, and oblique
images. (The transverse images, however, will not be distorted.)
The multienergy spatial registration, described in the previous chapter, is a measure of the cameras ability to maintain the same image magnification, regardless of the
energy of the x- or gamma rays forming the image. The multienergy spatial registration
is not only important in SPECT when imaging radionuclides such as Ga-67 and In-111,
which emit useful photons of more than one energy, but also because uniformity and
axis of rotation corrections, to be discussed shortly, determined with one radionuclide
will only be valid for others if the multienergy spatial registration is correct.
Alignment of Projection Images to the Axis of Rotation (COR

Calibration)
The axis of rotation (AOR) is an imaginary reference line about which the head or
heads of a SPECT camera rotate. If a radioactive line source were placed on the AOR,
each projection image would depict it as a vertical straight line near the center of
the image; this projection of the AOR into the image is called the center-of-rotation
(COR). The location of the COR in each projection image must be known to correctly
calculate the three-dimensional activity distribution from the projection images. Ideally, the COR is aligned with the center, in the x-direction, of each projection image.
However, there may be misalignment of the COR with the centers of the projection
images. This misalignment may be mechanical; for example, the camera head may
not be exactly centered in the gantry. It can also be electronic. The misalignment
may be the same amount in all projection images from a single camera head, or it
may vary with the angle of the projection image.
If a COR misalignment is not corrected, it causes a loss of spatial resolution in the
resultant transverse images. If the misalignment is large, it can cause a point source to
appear as a tiny doughnut (Fig. 19-10). (These doughnut artifacts are not seen in
clinical images; they are visible only in reconstructed images of point or line sources.
The doughnut artifacts caused by COR misalignment are not centered in the image
718
FIGURE 19-10 Center-of-rotation (COR) misalignment in SPECT. Small misalignments cause blurring (center),
whereas large misalignments cause point sources to appear as tiny doughnut artifacts (right). Such tiny
doughnut artifacts would only be visible in phantom studies and are unlikely to be seen in clinical images.
and so can be distinguished from the ring artifacts caused by nonuniformities.) Scintillation camera manufacturers provide software to assess and correct the effects of
COR misalignment. The COR alignment is assessed by placing a point source, several
point sources, or a line source in the cameras field of view, acquiring a set of projection images, and analyzing these images using the SPECT systems computer. If a line
source is used, it is placed parallel to the AOR.
The SPECT systems computer corrects the COR misalignment by shifting each
clinical projection image in the x-direction by the proper number of pixels prior to
filtered backprojection or iterative reconstruction. When a line source is used, the
COR correction can be performed separately for each transverse slice. If the COR misalignment varies with camera head angle, instead of being constant for all projection
images, it can only be corrected if the computer permits angle-by-angle corrections.
Separate assessments of the COR correction must be made for different collimators
and, on some systems, for different camera zoom factors and image formats (e.g., 642
versus 1282). The COR correction determined using one radionuclide will only be
valid for other radionuclides if the multienergy spatial registration is correct.
Uniformity
The uniformity of the camera head or heads is important; nonuniformities that are
not apparent in low count daily uniformity studies can cause significant artifacts in
SPECT. The artifact caused by a nonuniformity appears in transverse images as a ring
centered about the AOR (Fig. 19-11).
FIGURE 19-11 Image of a cylinder filled with

a uniform radionuclide solution, showing a ring
artifact due to a nonuniformity. The artifact is
the dark ring toward the center.
719
Multihead SPECT systems can produce partial ring artifacts when projection
images are not acquired by all heads over a 360-degree arc. Clinically, ring artifacts
are most apparent in high-count density studies, such as liver images. However, ring
artifacts may be most harmful in studies such as myocardial perfusion images in
which, due to poor counting statistics and large variations in count density, they may
not be recognized and thus lead to misinterpretation.
The causes of nonuniformities were discussed in the previous chapter. As mentioned in that chapter, modern scintillation cameras have digital circuits using lookup
tables to correct the X and Y position signals from each interaction for systematic
position-specific errors in event location assignment and the Z (energy) signal for
systematic position-specific variations in light collection efficiency. However, these
correction circuits cannot correct nonuniformity due to local variations in detection
efficiency, such as dents or manufacturing defects in the collimators.
If not too severe, nonuniformities of this latter type can be largely corrected. A
very high-count uniformity image is acquired. The ratio of the average pixel count to
the count in a specific pixel in this image serves as a correction factor for that pixel.
Following the acquisition of a projection image during a SPECT study, each pixel of
the projection image is multiplied by the appropriate correction factor before COR
correction and filtered backprojection or iterative reconstruction. For the high-count
uniformity image, at least 30 million counts should be collected for 642 pixel images
and 120 million counts for a 1282 pixel format. These high-count uniformity images
are typically acquired weekly to monthly. Correction images must be acquired for
each camera head and collimator. For cameras from some manufacturers, separate
correction images must be acquired for each radionuclide. The effectiveness of a cameras correction circuitry and of the use of high-count flood correction images can
be tested by acquiring a SPECT study of a large plastic container or a SPECT performance phantom filled with a well-mixed solution of Tc-99m and examining the
transverse images for ring artifacts. However, this testing will not assess parts of the
collimator or camera face outside the projected image of the container or phantom.
Camera Head Tilt

The camera head or heads must be aligned with the axis of rotation (AOR); for most
types of collimators, this requires that faces of the heads be exactly parallel to the AOR.
If they are not, a loss of spatial resolution and contrast will result from out-of-slice activity being backprojected into each transverse image slice, as shown in Figure 19-12.
d
ra
me
Camera head
AOR
hea
Ca
AOR
Patient
FIGURE 19-12 Head tilt. The camera head on the left is parallel to the axis of rotation (AOR), causing the counts
collected in a pixel of the projection image to be backprojected into the corresponding transverse image slice of
the patient. The camera head on the right is tilted, causing counts from activity outside of a transverse slice (along
the grey diagonal lines of response) to be backprojected into the transverse image slice (orange colored vertical
slice ).
720
The loss of resolution and contrast in each transverse image slice will be less toward
the center of the slice and greatest toward the edges of the image. If the AOR of the
camera is aligned to be level when the camera is installed and there is a flat surface on
the camera head that is parallel to the collimator face, a bubble level may be used to
test for head tilt. Some SPECT cameras require the head tilt to be manually adjusted
for each acquisition, whereas other systems set it automatically. The accuracy of the
automatic systems should be periodically tested. A more reliable method than the
bubble level is to place a point source in the camera FOV, centered in the axial (y)
direction, but near the edge of the field in the transverse (x) direction. A series of projection images is then acquired. If there is head tilt, the position of the point source
will vary in the y-direction from image to image. Proper head tilt can be evaluated
from cine viewing of the projection images.
SPECT Quality Control Phantoms

There are commercially available phantoms (Fig. 19-13) that may be filled with a
solution of Tc-99m or other radionuclide and used to evaluate system performance.
These phantoms are very useful for the semiquantitative assessment of spatial resolution, image contrast, and uniformity, although the small sizes of most such phantoms
allow uniformity to be assessed only over a relatively small portion of the cameras
face. They are used for acceptance testing of new systems and periodic testing, typically quarterly, thereafter. Table 19-1 provides a suggested schedule for a SPECT
quality control program.
19.3 Positron Emission Tomography

Positron emission tomography (PET) generates images depicting the distribution of
positron-emitting nuclides in patients. Nearly all PET systems manufactured today
are coupled to x-ray CT systems, with a single patient bed passing through the bores
of both systems, and are referred to as PET/CT systems. Figure 19-14 shows a
FIGURE 19-13 Jaszczak phantom for testing SPECT systems, a product of Data Spectrum Corporation.
(Photograph courtesy Data Spectrum Corporation.) A soluble radioactive material, typically labeled with Tc99m, is introduced into the phantom and mixed in the water until it is uniformly distributed. The acrylic plastic
spheres and rods are not radioactive and are cold objects in the radioactive solution.
721
TABLE 19-1RECOMMENDED SCHEDULE FOR ROUTINE QUALITY CONTROL
TESTING OF A SCINTILLATION CAMERA

TEST
FREQUENCY
NOTES
Set and check energy

discrimination window(s)
Before first use daily
Point (intrinsic) or planar

(extrinsic) source of radionuclide
Extrinsic or intrinsic low-count

uniformity images of all
camera heads
Before first use daily
510 million counts, depending

upon effective area of camera
head
Cine review of projection

images and/or review of
sinograma (S)
After each clinical SPECT

study
Check for patient motion
Visual inspection of collimators

for damage
Daily and when

changing collimators
If new damage found, acquire

a new high-count uniformity
calibration image
High count-density extrinsic or

intrinsic uniformity images of
all camera heads (S)
Weekly to monthly
30 million counts for 642 images

and 120 million counts for 1282
Spatial resolution check with

bar or hole pattern
Weekly
Center of rotation (S)
Weekly to monthly
Efficiency of each camera head
Quarterly
Reconstructed cylindrical
phantom uniformity (S)
Monthly to quarterly
Cylindrical phantom filled with

Tc-99m solution
Point source reconstructed

spatial resolution (S)
Quarterly, semiannually,
or annually
Point source
Point or line source(s), as

recommended by manufacturer
Reconstructed SPECT phantom (S) Quarterly
Using a phantom such as the

one shown in Figure19-13
Pixel size check
Quarterly
Two point sources
Head tilt angle check (S)
Quarterly
Bubble level or point source
Extrinsic uniformity images

of all collimators not tested
above
Quarterly or
semiannually
Planar source. High-count

density images of all collimators
used for SPECT
Multienergy spatial
registration
Quarterly, semiannually,
or annually
Ga-67 point source
Count rate performance
Annually
Tc-99m source
The results of these tests are to be compared with baseline values, typically determined during acceptance
testing. If the manufacturer recommends or an accrediting body specifies additional tests or more frequent
testing, these recommendations or specifications should take precedence. For tests with multiple frequencies
listed, it is recommended that the tests be performed initially at the higher frequency, but the frequency be
reduced if the measured parameters prove stable. Tests labeled (S) need not be performed for cameras used
only for planar imaging.
a
A sinogram is an image containing projection data corresponding to a single transaxial image of the patient.
Each row of pixels in the sinogram is the row, corresponding to that transaxial image, of one projection image.
PET/CT system. This section discusses PET imaging systems; their use in PET/CT
systems is discussed later in this chapter.
In a typical PET system, several rings of detectors surround the patient. PET
scanners use annihilation coincidence detection (ACD) instead of collimation to obtain
projections of the activity distribution in the subject. The PET systems computer
then reconstructs the transverse images from the projection data, as does the computer of an x-ray CT or SPECT system. Modern PET scanners are multislice devices,
722
FIGURE 19-14 A commercial PET/CT scanner. (Photograph courtesy Siemens Medical

Solutions, Nuclear Medicine Group.)
permitting the simultaneous acquisition of as many as 109 transverse images over

21.6 cm of axial distance. The clinical importance of PET today is largely due to its
ability to image the radiopharmaceutical fluorine-18 fluorodeoxyglucose (FDG), a
glucose analog used for differentiating malignant neoplasms from benign lesions,
staging patients with malignant neoplasms, monitoring the response to therapy for
neoplasms, differentiating severely hypoperfused but viable myocardium from scar,
and other applications. A few other positron-emitting radiopharmaceuticals have
been approved for use and many are under development or clinical investigation.

Annihilation Coincidence Detection
Positron emission is a mode of radioactive transformation and was discussed in
Chapter 15. Positrons emitted in matter lose most of their kinetic energy by causing
ionization and excitation. When a positron has lost most of its kinetic energy, it interacts with an electron by annihilation, as shown on the left in Figure 19-15. The entire
mass of the electron-positron pair is converted into two 511-keV photons, which are
emitted in nearly opposite directions. In solids and liquids, positrons travel only very
short distances (see Table 19-3) before annihilation.
If both photons from an annihilation interact with detectors and neither photon is scattered in the patient, the annihilation occurred near the line connecting the
two interactions, as shown on the right in Figure 19-15. Circuitry within the scanner identifies pairs of interactions occurring at nearly the same time, a process called
annihilation coincidence detection (ACD). The circuitry of the scanner then determines the line in space connecting the locations of the two interactions. Thus, ACD
establishes the trajectories of detected photons, a function performed by collimation in
723
511 keV
photon
e
+
Nucleus emits
positron
Annihilation
of positron
and electron
511 keV
photon
FIGURE 19-15 Annihilation coincidence detection (ACD). When a positron is emitted by a nuclear transformation, it scatters through matter losing energy. After it loses most of its energy, it annihilates with an electron, resulting in two 511-keV photons that are emitted in nearly opposite directions (left). When two interactions are nearly
simultaneously detected within a ring of detectors surrounding the patient (right), it is presumed that an annihilation
occurred on the line connecting the interactions (i.e., line of response). Thus, ACD, by determining the path of the
detected photons, performs the same function for a PET scanner as does the collimator of a scintillation camera.
SPECT systems. However, the ACD method is much less wasteful of photons than
collimation. Additionally, ACD avoids the degradation of spatial resolution with
distance from the detector that occurs when collimation is used to form projection
images.
True, Random, and Scatter Coincidences

A true coincidence is the nearly simultaneous interaction with the detectors of emissions resulting from a single nuclear transformation. A random coincidence (also called
an accidental or chance coincidence), which mimics a true coincidence, occurs when
emissions from different nuclear transformations interact nearly simultaneously with
the detectors (Fig. 19-16). A scatter coincidence occurs when one or both of the photons from a single annihilation are scattered, and both are detected (Fig. 19-16).
A scatter coincidence is a true coincidence, because both interactions result from a
single positron annihilation. Random coincidences and scatter coincidences result in
misplaced coincidences, because they are assigned to lines of response (LORs) that
do not intersect the actual locations of the annihilations. They are therefore sources of
noise, whose main effects are to reduce image contrast and increase statistical noise.
Detection of Interactions
Scintillation crystals coupled to photomultiplier tubes (PMTs) are used as detectors in
all commercial PET systems today; the low intrinsic efficiencies of gas-filled and semiconductor detectors for detecting 511-keV photons make them impractical for use
FIGURE 19-16 True coincidence (left), scatter coincidence (center), and random (accidental) coincidence
(right). A scatter coincidence is a true coincidence, because it is caused by a single nuclear transformation,
but results in a count attributed to the wrong LOR (dashed line). The random coincidence is also attributed to
the wrong LOR.
724
in PET. The signals from the PMTs are processed using pulse mode (the signals from
each interaction are processed separately from those of other interactions) to create
signals identifying the position in the detector, deposited energy, and time of each
interaction. The energy signal is used for energy discrimination to reduce mispositioned events due to scatter and the time signal is used for coincidence detection and,
in some PET systems, time-of-flight determination (discussed later in this chapter).
In early PET scanners, each scintillation crystal was coupled to a single PMT. In this
design, the size of the individual crystal largely determined the spatial resolution of the
system; reducing the size (and therefore increasing the number of crystals) improved
the resolution. It became increasingly costly and impractical to pack more and more
smaller PMTs into each detector ring. Modern designs couple larger crystals to more
than one PMT; one such design is shown in Figure 19-17. The relative magnitudes of
the signals from the PMTs coupled to a single crystal are used to determine the position
of the interaction in the crystal, as in a scintillation camera. In such a system, the crystal
and four PMTs comprise a detector. However, an interaction is localized to one of the
portions of the crystal separated by the slits and thus each such segment of the crystal
can be considered to be a detector element from the perspective of image formation.
The scintillation material must emit light very promptly to permit true coincident
interactions to be distinguished from random coincidences and to minimize deadtime count losses at high interaction rates. Also, to maximize the counting efficiency,
the material must have a high linear attenuation coefficient for 511-keV photons. A
high conversion efficiency (fraction of deposited energy emitted as light or UV radiation) is also important; it permits more precise event localization in the detectors and
better energy discrimination. Until recently, most PET systems used crystals of bismuth germanate (Bi4Ge3O12, abbreviated BGO). The light output of BGO is only 12%
to 14% of that of NaI(Tl), but its greater density and average atomic number give it
a much higher efficiency in detecting 511-keV annihilation photons. Light is emitted
rather slowly from BGO (decay constant of 300 ns), which contributes to dead-time
count losses and random coincidences at high interaction rates. Several inorganic
scintillators that emit light more quickly are replacing BGO. These include lutetium
2 by 2 array
of PMTs
Slits cut into

crystal
Scintillation crystal
FIGURE 19-17 A technique for coupling scintillation crystals to photomultiplier tubes (PMTs). The relative heights of the pulses from the four PMTs are used to determine the position of each interaction in the
crystal. The thick (2 to 3 cm) crystal is necessary to provide a reasonable detection efficiency for the 511-keV
annihilation photons. The slits cut in the scintillation crystal form light pipes, limiting the spread of light from
interactions in the front portion of the crystal, which otherwise would reduce the spatial resolution. This design
permits four PMTs to serve 64 detector elements.
725
oxyorthosilicate (Lu2SiO4O, abbreviated LSO), lutetium yttrium oxyorthosilicate

(LuxY2-xSiO4O, abbreviated LYSO), and gadolinium oxyorthosilicate (Gd2SiO4O,
abbreviated GSO), all activated with cerium. Their attenuation properties are nearly
as favorable as those of BGO and their much faster light emission produces better
performance at high interaction rates, especially in reducing dead-time effects and in
discriminating between true and random coincidences. A disadvantage of LSO and
LYSO is that they are slightly radioactive; about 2.6% of naturally occurring lutetium
is radioactive Lu-176, which has a half-life of 38 billion years, producing about 295
nuclear transformations per second in each cubic centimeter of LSO. NaI(Tl) was
used as the scintillator in some less-expensive PET systems. The properties of BGO,
LSO, LYSO, and GSO are contrasted with those of NaI(Tl) in Table 19-2.
The energy signals from the detectors are sent to energy discrimination circuits,
which can reject events in which the deposited energy differs significantly from 511
keV to reduce the effect of photon scattering in the patient. However, some annihilation photons that have escaped the patient without scattering interact with the
detectors by Compton scattering, depositing less than 511 keV. An energy discrimination window that encompasses only the photopeak rejects these valid interactions
as well as photons that have scattered in the patient. The energy window can be set
to encompass only the photopeak, with maximal rejection of scatter, but also reducing the number of valid interactions detected, or the window can include part of the
Compton continuum, increasing the sensitivity, but also increasing the number of
scattered photons detected.
The time signals of interactions not rejected by the energy discrimination circuits are used for coincidence detection. When a coincidence is detected, the circuitry or a computer in the scanner determines a line in space connecting the two
interactions, as shown in Figure 19-15. This is called a line of response. The number
of coincidences detected along each LOR is stored in the memory of the computer.
Figure 19-18 compares the acquisition of projection data by SPECT and PET systems. Once data acquisition is complete, the computer uses the projection data to
produce transverse images of the radionuclide distribution in the patient, as in x-ray
CT or SPECT.
TABLE 19-2PROPERTIES OF SEVERAL INORGANIC SCINTILLATORS
OFINTEREST IN PET
DECAY
SCINTILLATOR CONSTANT (ns)
CONVERSION
PEAK
ATTENUATION EFFICIENCY
WAVELENGTH ATOMIC
DENSITY COEFFICIENT
RELATIVE TO
(nm)
NUMBERS (g/cm3)
511 keV (cm1) NaI
NaI(Tl)
250
415
11,53
3.67
0.343
100%
BGO
300
460
83,32,8
7.17
0.964
12%14%
GSO(Ce)
56
430
64,14,8
6.71
0.704
41%
LSO(Ce)
40
420
71,14,8
7.4
0.870
75%
Decay constants, peak wavelengths, densities, and conversion efficiencies of BGO, GSO, and LSO (From Ficke
DC, Hood JT, Ter-Pogossian MM. A spheroid positron emission tomograph for brain imaging: a feasibility
study. J Nucl Med 1996;37:1222.)
LYSO(Ce) has a decay constant, peak emission wavelength, and relative conversion efficiency similar to those
of LSO (Ce), but its density and attenuation coefficient are less. Its density and attenuation coefficient vary
with the proportion of yttrium to lutetium. (Information on LYSO from Chen J, Zhang L, Zhu RY. Large size
LYSO crystals for future high energy physics experiments. IEEE Trans Nucl Sci 2005;52(6):31333140.)
726
T ad
EC he
SP era
m
ca
FIGURE 19-18 Acquisition

of projection data by a SPECT
system (left) and a PET system
(right), showing how each system collects data for two projections. However, the PET system
collects data for all projections
simultaneously, whereas the
scintillation camera head of the
SPECT system must move to
collect data from each projection angle.
SPECT
camera head
Timing of Interactions and Detection of Coincidences

To detect coincidences, the times of individual interactions in the detectors must be
compared. However, interactions themselves cannot be directly detected; instead,
the signals caused by interactions are detected. In scintillators, the emission of light
after an interaction is characterized by a relatively rapid increase in intensity followed by a more gradual decrease. A timing circuit connected to each detector must
designate one moment that follows each interaction by a constant time interval. The
time signal is determined from the leading edge of the electrical signal from the PMTs
connected to a detector crystal because the leading edge is steeper than the trailing
edge. When the time signals from two detectors occur within a selected time interval
called the time window, a coincidence is recorded. A typical time window for a system
with BGO detectors is 12 ns. A typical time window for a system with LSO detectors,
which emit light more promptly, is 4.5 ns.
True versus Random Coincidences

The rate of random coincidences between any pair of detectors is:

Rrandom t S1S2
[19-1]
where is the coincidence time window and S1 and S2 are the actual count-rates
of the detectors, often called singles rates. The time window is the designated time
interval during which a pair of interactions in different detectors is considered to
be a coincidence. The ratio of random to true coincidences increases with increasing activity in the subject and decreases as the time window is shortened. However, there is a limit to how small the time window can be. The time window must
accommodate the difference in arrival times of true coincidence photons from annihilations occurring in locations up to the edge of the devices FOV. For example,
if the longest path for an unscattered annihilation photon crossing the FOV is
70 cm, the time window can be no shorter than
t 0.7 m/(3.0 108 m/s) 2.33 ns
where 3.0 108 m/s is the speed of light. Furthermore, the time window must also
be sufficiently long to accommodate the imprecision in the measured times of interactions. Scintillation materials that emit light promptly permit the use of shorter time
windows and thus better discrimination between true and random coincidences.
Because the singles rates in individual detectors, ignoring dead-time count losses,
are approximately proportional to the activity in the patient, Equation 19-1 shows
that the rate of random coincidences is approximately proportional to the square
of the activity in the patient. The true coincidence rate, ignoring dead-time count
727
losses, is approximately proportional to the activity in the patient and so the ratio
of random coincidences to true coincidences is approximately proportional to the
administered activity.
Although individual random coincidences cannot be distinguished from true
coincidences, methods of correction are available. The number of random coincidences along each LOR can be measured by adding a time delay to one of the two timing signals used for coincidence detection, so that no true coincidences are detected.
Alternatively, the number of random coincidences may be calculated from the single
event count-rates of the detectors, using Equation 19-1. The number of random coincidences is then subtracted from the number of true plus random coincidences for
each LOR. This method corrects for the magnitude of signal from random coincidences, but not for the statistical noise that they introduce.
Scatter Coincidences
As previously mentioned, a scatter coincidence occurs when one or both of the photons
from an annihilation are scattered in the patient and both are detected (Fig.19-16).
The fraction of scatter coincidences is dependent upon the amount of scattering material and thus is less in head than in body imaging. Because scatter coincidences are
true coincidences, reducing the activity administered to the patient, reducing the time
window, or using a scintillator with faster light emission does not significantly reduce
the scatter coincidence fraction. Furthermore, the corrections for random coincidences
do not compensate for scatter coincidences. The energy discrimination circuits of the
PET scanner can be used to reject some scatter coincidences. As was mentioned previously, energy discrimination to reduce scatter is less effective in PET than in SPECT,
because many of the 511-keV annihilation photons interact with the detectors by
Compton scattering, depositing less than their entire energies in the detectors. Using
an energy window closely enveloping the photopeak will reject many annihilation
photons that have not scattered in the patient. The fraction of scatter coincidences
can also be reduced by two-dimensional data acquisition and axial septa, as described
in the following section. Approximation methods are available for scatter correction.
These methods correct for the magnitude of the signal from scatter coincidences, but
not for the statistical noise introduced by scatter coincidences.
Two- and Three-Dimensional Data Acquisition

In two-dimensional (slice) data acquisition, coincidences are detected and recorded
only within each ring of detector elements or within a few adjacent rings of detector
elements. PET scanners designed for two-dimensional acquisition have axial septa,
thin annular collimators, typically made of tungsten, to prevent most radiation emitted by activity outside a transaxial slice from reaching the detector ring for that slice
(Fig. 19-19). The fraction of scatter coincidences is greatly reduced in PET systems
using two-dimensional data acquisition and axial collimation because of the geometry. Consider an annihilation occurring within a particular detector ring with the
initial trajectories of the annihilation photons toward the detectors. If either of the
photons scatters, it is likely that the new trajectory of the photon will cause it to miss
the detector ring, thereby preventing a scatter coincidence. Furthermore, most photons from out-of-slice activity are absorbed by the axial septa.
In two-dimensional data acquisition, coincidences within one or more pairs
of adjacent detector rings may be added to improve the sensitivity, as shown in
Figure 19-20. The data from each pair of detector rings are added to that of the
slice midway between the two rings. For example, if N is the number of a particular ring of detector elements, coincidences between rings N1 and N1 are
728

FIGURE 19-19 Side view of PET scanner
illustrating two-dimensional data acquisition. The axial septa prevent photons from
activity outside the field-of-view (A) and
most scattered photons (B) from causing
counts in the detectors. However, many
valid photon pairs (C) are also absorbed.
Detector elements
Axial septa
A
B
added to the projection data for ring N. For further sensitivity, coincidences between
rings N2 and N2 can also be added to those of ring N. After the data from these
pairs of detector rings are added, a standard two-dimensional reconstruction is performed to create a transverse image for this slice. Similarly, coincidences between two
immediately adjacent detector rings, that is, between rings N and N1, can be used
to generate a transverse image halfway between these rings. For greater sensitivity,
coincidences between rings N1 and N2 can be added to these data. However,
increasing the number of pairs of adjacent rings used in two-dimensional acquisition
reduces the axial spatial resolution.
In three-dimensional (volume) data acquisition, axial septa are not used and
coincidences are detected between many or all detector rings (Fig. 19-21). Threedimensional acquisition greatly increases the number of true coincidences detected
and may permit smaller activities to be administered to patients, compared to twodimensional image acquisition. There are disadvantages to three-dimensional data
FIGURE 19-20 Side view of PET scanner performing

two-dimensional data acquisition, showing LORs for a
single transverse image. Cross-ring coincidences have
been added to those occurring within the ring of detector elements. This increases the number of coincidences
detected, but causes a loss of axial spatial resolution that
increases with distance from the axis of the scanner.
Detector elements
Axial septa
Detector elements
Axial septa
removed
A
B
729
FIGURE 19-21 Side view of PET scanner

illustrating three-dimensional data acquisition. Without axial septa, interactions from
activity outside the FOV (A) and scattered
photons (B) are greatly increased, increasing the dead time, random coincidence
fraction, and scatter coincidence fraction.
However, the number of true coincidences
(C) detected is also greatly increased.
acquisition. For the same administered activity, the greatly increased interaction rate
increases the random coincidence fraction and the dead-time count losses. Furthermore, the scatter coincidence fraction is much larger and the number of interactions
from activity outside the FOV is greatly increased. (Activity outside the FOV causes
few true coincidences, but increases the rate of random coincidences detected and
dead-time count losses.) Some PET systems are equipped with retractable axial septa,
permitting them to perform two- or three-dimensional acquisition.
Figure 19-22 shows the efficiency of coincidence detection when two- or threedimensional acquisition is used. In two-dimensional acquisition, the efficiency is
nearly constant along the axial length of the detector rings. In three-dimensional
Coincidence
Detection
Efficiency
3D
2D
Position Along Axis of PET

FIGURE 19-22 Efficiency of coincidence detection along axis of detector rings for two- and three-dimensional
data acquisition. The three-dimensional acquisition example presumes that coincidences are detected among
all rings of detector elements. The coincidence detection efficiency in two-dimensional acquisition is least if
coincidences are detected only within individual rings of detector elements; it substantially increases when
coincidences are also detected among adjacent rings, as shown in Figure 19-20.
730
acquisition, if coincidences are detected among all rings of detector elements, the
coincidence detection efficiency increases linearly from the ends of the rings to
the center. PET scans of extended lengths of patients are commonly accomplished
by a discontinuous motion of the patient table through the system, with the table
stopping for individual bed positions. Because of the greatly varying coincidence
detection efficiency, much greater overlap of bed positions is necessary when using
three-dimensional acquisitions.
Transverse Image Reconstruction

After the projection data are acquired, the data for each LOR are corrected for random coincidences, scatter coincidences, dead-time count losses, and attenuation
(described below). Following these corrections, transverse image reconstruction is
performed. For two-dimensional data acquisition, image reconstruction methods are
similar to those used in SPECT. As in SPECT, either filtered backprojection or iterative reconstruction methods can be used. For three-dimensional data acquisition,
special three-dimensional analytical or iterative reconstruction methods are required.
These are beyond the scope of this chapter, but are discussed in the suggested readings listed at the end of the chapter.
An advantage of PET over SPECT is that, in PET, the correction for nonuniform
attenuation can be applied to the projection data before reconstruction. In SPECT,
the correction for nonuniform attenuation is intertwined with and complicates the
reconstruction process.
Time of Flight Determination

In theory, measurement of the difference between the times of the interactions in
the detectors of a pair of annihilation photons would determine the point on a line
between the two interactions where an annihilation occurred. If this were possible,
PET would be inherently tomographic and the reconstruction of transverse images
from projection data would not be necessary.
The speed of light is 3.0 108 m/s. To determine the location of an annihilation
within 1 cm would require being able to measure a difference in interaction times of
t 1 cm / 3.0 1010 cm/s 0.033 ns. Such precision in timing is not possible
with current PET scintillators. Fast scintillators such as LSO and LYSO permit determination of the location of an annihilation within several cm. This information can
be used in the tomographic reconstruction process to improve the signal-to-noise
ratio in the images.
Performance
Spatial Resolution
Modern whole-body PET systems achieve a spatial resolution slightly better than
5-mm FWHM of the LSF in the center of the detector ring when measured by the
NEMA standard, Performance Measurements of Positron Emission Tomographs (National
Electrical Manufacturers Association, NEMA NU 2-2007, Performance Measurements
of Positron Emission Tomographs). There are three factors that primarily limit the
spatial resolution of PET scanners: (1) the intrinsic spatial resolution of the detectors,
(2) the distances traveled by the positrons before annihilation, and (3) the fact that
the annihilation photons are not emitted in exactly opposite directions from each
other. The intrinsic resolution of the detectors is the major factor determining the
spatial resolution in current scanners. In older systems in which the detectors consisted of separate crystals each attached to a single PMT, the size of the crystals
FIGURE 19-23 Cause of reduced spatial resolution with

distance from center of PET scanner. Only three crystal segments in the detector ring are shown. For coincident interactions in detectors A and B, there is little uncertainty in the LOR.
When coincidences occur in detectors B and C, there is greater
uncertainty in the LOR. Using thinner crystals would reduce
this effect, but would also reduce their intrinsic detection efficiency and thus the number of coincidences detected.
crystal thickness
731
etermined the resolution. In clinical imaging, spatial resolution is also reduced by

d
organ motion, most commonly due to respiration.
The spatial resolution of a PET system is best in the center of the detector ring
and decreases slightly (FWHM increases) with distance from the center. This occurs
because of the considerable thickness of the detectors. Uncertainty in the depth of
interaction causes uncertainty in the LOR for annihilation photons that strike the
detectors obliquely. Photons emitted from the center of the detector ring can only
strike the detectors head-on, but many of the photons emitted from activity away
from the center strike the detectors from oblique angles (Fig. 19-23). Some newer
PET systems can estimate the depths of interactions in the detectors and have more
uniform spatial resolution across a transverse image.
The distance traveled by the positron before annihilation also slightly degrades
the spatial resolution. This distance is determined by the maximal positron energy
of the radionuclide and the density of the tissue. A radionuclide that emits lower
energy positrons yields better resolution. Table 19-3 lists the maximal energies of
positrons emitted by radionuclides commonly used in PET. Activity in denser tissue
yields higher resolution than activity in less dense tissue such as lung tissue.
TABLE 19-3PROPERTIES OF POSITRON-EMITTING RADIONUCLIDES
COMMONLY USED IN PET
NUCLIDE
HALFLIFE (min)
POSITRONS PER
TRANSFORMATION
MAXIMAL
POSITRON
ENERGY (keV)
MAXIMAL
POSITRON
RANGE (mm)a
C-11
20.4
1.00
960
4.2
N-13
10.0
1.00
1198
5.4
O-15
2.0
1.00
1732
8.4
F-18
110
0.97
634
2.4
Most common
use is F-18
fluorodeoxyglucose.
Rb-82
1.3
0.95
3356
17
Produced by a Sr-82/
Rb-82 generator.
NOTES
The average positron range is much less than the maximal range, because positrons are emitted with a
continuum of energies and because the paths of positrons in matter are not straight.
a
In water from ICRU Report 37.
732
Although positrons lose nearly all of their momentum before annihilation, the
positron and electron possess some residual momentum when they annihilate. Conservation of momentum predicts that the resultant photons will not be emitted in
exactly opposite directions. This causes a small loss of resolution, which increases
with the diameter of the detector ring.
Efficiency in Annihilation Coincidence Detection

Consider a point source of a positron-emitting radionuclide in air midway between
two identical detectors (Fig. 19-24) and assume that all positrons annihilate within
the source. The true coincidence rate of the pair of detectors is

RT 2 AG 2
[19-2]
where A is the rate of positron emission by the source, G is the geometric efficiency
of either detector, and is the intrinsic efficiency of either detector. (Geometric and
intrinsic efficiency were defined in Chapter 17.) Because the rate of true coincidences
detected is proportional to the square of the intrinsic efficiency, maximizing the
intrinsic efficiency is very important in PET. For example, if the intrinsic efficiency
of a single detector for a 511-keV photon is 0.9, 81% of the annihilation photon
pairs emitted toward the detectors will have coincident interactions. However, if the
intrinsic efficiency of a single detector is 0.1, only 1% of pairs emitted toward the
detectors will have coincident interactions.
As mentioned previously, most PET systems today use crystals of high density,
high atomic number scintillators such as BGO, LSO, LYSO, or GSO. These crystals
are typically about two to three cm thick. For example, a 2-cm thick crystal of LSO
has an intrinsic efficiency of 82%, although energy discrimination reduces the fraction of detected photons used for coincidence detection below this value.
As mentioned above in Two- and Three-Dimensional Data Acquisition, increasing the axial acceptance angle of annihilation photons greatly increases the efficiency.
In PET systems with three-dimensional data acquisition, this can be augmented by
adding additional rings of detectors.
Attenuation and Attenuation Correction in PET

Attenuation in PET differs from attenuation in SPECT, because both annihilation
photons must escape the patient to cause a coincident event to be registered. The
probability of both photons escaping the patient without interaction is the product
of the probabilities of each escaping:

(e-mx ) (e -m ( d - x ) ) = e-md
[19-3]
where d is the total path length through the patient, x is the distance one photon must
travel to escape, and (d-x) is the distance the other must travel to escape (Fig. 19-25).
Thus, the probability of both escaping the patient without interaction is independent
of where on the line the annihilation occurred and is the same as the probability of
FIGURE 19-24 Point source of a positron-emitting radionuclide, in air, midway between two identical detectors.
Source containing
positron-emitter
Detector
Detector
Annihilation
photon
733
FIGURE 19-25 Attenuation in PET.

The probability that both annihilation
photons emitted along a particular
LOR escape interaction in the patient
is independent of the position on the
LOR where the annihilation occurred.
d
d-x
Patient
X
Annihilation
photon
a single 511-keV photon passing entirely through the patient along the same path.
(Equation 19-3 was derived for the case of uniform attenuation, but this principle is
also valid for nonuniform attenuation.)
Even though the attenuation coefficient for 511-keV annihilation photons in soft
tissue (m/r 0.095 cm2/g) is lower than those of photons emitted by most radionuclides used in SPECT (m/r 0.15 cm2/g for 140 keV gamma rays), the average path
length for both to escape is much longer. For a 20-cm path in soft tissue, the chance
of both annihilation photons of a pair escaping the tissue without interaction is only
about 15%. Thus, attenuation is more severe in PET than in SPECT. The vast majority of the interactions with tissue are Compton scattering. Attenuation causes a loss
of information and, because the loss is not the same for all LORs, causes artifacts in
the reconstructed tomographic images. The loss of information also contributes to
the statistical noise in the images.
As in SPECT, both approximate methods and methods using radioactive sources
to measure the attenuation have been used for attenuation correction in PET. Most
of the approximate methods used a profile of the patient and presume a uniform
attenuation coefficient within the profile.
Many older PET systems provided one or more retractable positron-emitting
sources inside the detector ring between the detectors and the patient to measure
the transmission of annihilation photons from the sources through the patient. As
mentioned above, the probability that a single annihilation photon from a source
will pass through the patient without interaction is the same as the probability that
both photons from an annihilation in the patient traveling along the same path
through the patient will escape interaction. These sources were usually configured as rods and were parallel to the axis of the scanner (Fig. 19-26). The sources
revolved around the patient, so that attenuation was measured along all LORs
through the patient. These sources usually contained Ge-68, which has a halflife of 271 days and decays to Ga-68, which primarily decays by positron emission. Alternatively, a source containing a gamma ray-emitting radionuclide, such
as Cs-137 (662-keV gamma ray), has been used for attenuation measurements
instead of a positron-emitting radionuclide source. When a gamma ray-emitting
radionuclide is used to measure the attenuation, the known position of the source
734

FIGURE 19-26 Rod source for attenuation correction in a dedicated PET system.
Rod
source
and the location where a gamma ray is detected together determine a LOR for
that interaction. In either case, the transmission data acquired using the external
sources were used to correct the emission data for attenuation prior to transverse
image reconstruction.
There are the advantages and disadvantages to attenuation correction from transmission measurements using radioactive sources. The transmission measurements
increased the imaging time; slightly increased the radiation dose to the patient; and
increased the statistical noise in the images. Also, spatial misregistration between the
attenuation map and the emission data, most often caused by patient motion, can
cause significant artifacts if the transmission data are not acquired simultaneously
with the emission data.
Today, nearly every PET system manufactured is coupled to an x-ray CT system.
These are called PET/CT systems. On such systems, the x-ray transmission information is used to correct the PET information for attenuation and so they are not
provided with radioactive transmission sources for attenuation correction. PET/CT
systems are discussed in detail later in this chapter.
Alternatives to Dedicated PET SystemsMultihead SPECT

Cameras with Coincidence Detection Capability and SPECT
with High Energy Collimators
Double and triple head SPECT cameras with coincidence detection capability have
been commercially available. These cameras had planar NaI(Tl) crystals coupled to
PMTs. With standard nuclear medicine collimators mounted, they functioned as
conventional scintillation cameras and could perform planar nuclear medicine imaging
and SPECT. With the collimators removed or replaced by slit collimators which reduce
the rate of interactions by photons from out-of-field activity, they could perform coincidence imaging of positron-emitting radiopharmaceuticals. When used for coincidence
imaging, these cameras provided spatial resolution similar to that of dedicated PET
systems. The smaller attenuation coefficient of sodium iodide with respect to common PET scintillators and the thin crystal of a typical scintillation camera limited the
intrinsic efficiency of each detector in detecting annihilation photons. For this reason,
many cameras intended for coincidence imaging had thicker NaI crystals than standard scintillation cameras, with thicknesses ranging from 1.27 to 2.5 cm (1/2 to 1
inch). These thicker crystals significantly increased the number of true coincidences
detected. However, even with thicker crystals, they detected much smaller numbers
of true coincidences than dedicated PET systems. The lower coincidence detection
sensitivity caused more statistical noise in the images than in images from a dedicated
735
PET system. Many studies were performed comparing multihead scintillation cameras
capable of coincidence detection with dedicated PET systems for the task of detecting
lesions with concentrations of F-18 FDG; these studies showed that the coincidence
scintillation cameras had comparable performance for the detection of large lesions, but
significantly worse performance in detecting small lesions in the patients, in comparison to dedicated PET systems (Boren et al., 1999; Delbeke et al., 1999; Oturai et al.,
2004; Tiepolt et al., 2000; Yutani et al., 1999; Zhang et al., 2002; Zimny et al., 1999).
Some scintillation cameras can perform planar imaging and SPECT of positron
emitters using collimators designed for 511-keV photons. These collimators have very
thick septa and have very low collimator efficiencies. Despite the thick septa, the images
produced suffer from considerable septal penetration. Also, the thick septa produce
collimator pattern artifacts in the images. However, attenuation by the patient is less
significant than in coincidence imaging. The gantries of such cameras must be able to
support very heavy camera heads, because of the weight of 511-keV collimators and
because the camera heads must contain adequate shielding for 511-keV photons from
outside the FOV. F-18 FDG imaging with high-energy collimators does not appear to
be acceptable for the evaluation and staging of neoplasms (Macfarlane et al., 1995),
because of its limited detection efficiency and spatial resolution.
Quality Control in PET

Daily, typically before the first patient imaging, a scan is performed of a uniform
positron-emitting source. This may be a line source or a cylindrical source. Ge-68 is a
radionuclide commonly used in these sources; its daughter Ga-68, with which it is in
secular equilibrium, emits positrons. This scan, typically displayed as detector maps or
sinograms, will reveal poorly performing or inoperative detectors, detectors requiring
normalization, and detectors with improperly set energy windows. Tomographic uniformity is periodically assessed using a cylindrical phantom with the uniform distribution of a positron-emitting radionuclide. If the daily or periodic uniformity scans reveal
changes in detector efficiencies, a detector normalization calibration is performed to
measure the efficiencies of all the detector LORs in the system and update the stored
normalization factors. Periodically, an absolute activity calibration should also be performed if clinical measurements of activity concentration are to be performed. A medical physicist should perform a complete systems test and review the quality control
program at least annually. This systems test assesses many parameters, including spatial
resolution, statistical noise, count-rate performance, sensitivity, and image quality.
19.4 Dual Modality ImagingSPECT/CT, PET/CT, and PET/MRI

Nuclear medicine tomographic imaging, whether SPECT or PET, of cancer, infections, and inflammatory diseases often reveals lesions with greater radionuclide concentrations than surrounding tissue, but often provides little information regarding
their exact locations in the organs of the patients. Spatial coregistration of the SPECT
or PET images with images from another modality, such as x-ray CT or MRI, that
provides good depiction of anatomy can be very useful. This can be accomplished by
imaging the patient using separate SPECT or PET and CT or MRI systems and aligning the resultant images using multimodality image registration software. However,
it is difficult to align the patient on the bed of the second imaging system exactly the
same as on the first system, which in turn makes it very difficult to accurately align
the SPECT or PET data with the CT or MRI data.
736
As mentioned earlier in this chapter, another problem common to both SPECT

and PET is attenuation of the gamma rays, x-rays, or annihilation photons by the
patient, causing a reduction in apparent activity toward the centers of transverse
images and attenuation artifacts, particularly when attenuation is nonuniform.
In both SPECT and PET, manufacturers have provided the capability to perform
attenuation correction using sealed radioactive sources to measure the transmission
through the patient from the various projection angles. As mentioned earlier, the use
of radioactive sources for attenuation correction in SPECT was not met with universal acceptance, whereas the use of external sources for attenuation correction in PET
was generally accepted. However, in PET, the use of radioactive sources for attenuation correction increased imaging times, reducing patient throughput.
To solve these problems, systems have been developed that incorporate a PET or
SPECT system and a conventional x-ray CT system in a single gantry or in coupled
gantries (Fig. 19-14 and Fig. 19-1A,B). The patient lies on a bed that passes through
the bores of both systems. X-ray CT imaging is performed of the same length of the
patient as the PET or SPECT imaging, either before or after the PET or SPECT image
acquisition. Because the patients position on the bed is the same during CT and PET
or SPECT image acquisition, accurate coregistration of the CT and PET or SPECT
information is usually possible. In the coregistered images, the PET or SPECT information is usually superimposed in color on grayscale CT images (Fig. 19-27).
FIGURE 19-27 Coregistered PET and CT images. (Images courtesy of D. K. Shelton, M.D., UC Davis Medical
Center.)
737
Advantages of using x-ray CT systems instead of radioactive sources for attenuation

correction are much quicker attenuation scans, permitting increased patient throughput
and reducing the likelihood of patient motion; less statistical noise in the attenuation
data due to the high x-ray flux from the CT; and higher spatial resolution attenuation
data. However, there are disadvantages as well, which are discussed below.
Manufacturers have provided SPECT/CT and PET/CT systems with x-ray CT
subsystems that are capable of producing diagnostic quality CT images and with
less-expensive CT systems that are not capable of producing diagnostic quality CT
images. If the x-ray CT subsystem is capable of diagnostic quality imaging, it may be
used with low tube current, providing low dose but nondiagnostic CT information
for attenuation correction and anatomic localization only, or with high tube current,
providing diagnostic quality CT images.
Nearly all PET systems sold today are integrated PET/x-ray CT systems, because
of their ability to provide attenuation-corrected PET images coregistered to x-ray CT
images with high patient throughput. Efforts are underway to develop combined
PET/MRI systems. These are discussed later in this chapter.
Attenuation Correction in PET/CT

As mentioned above, the x-ray CT provides information that can be used for attenuation correction of the PET information, replacing the positron-emitting rod sources
formerly used for attenuation correction by PET systems. A modern x-ray CT system
can provide the attenuation information, with very little statistical noise, in a fraction
of a minute, in comparison with the many minutes required when positron-emitting
rod sources were used.
An x-ray CT system measures the average linear attenuation coefficients, averaged over the x-ray energy spectrum, for individual volume elements (voxels) of the
patient. However, linear attenuation coefficients for 511-keV annihilation photons
are necessary for attenuation correction of PET information. Linear attenuation coefficients, measured using the x-rays from x-ray CT systems, are very different from
those for 511-keV photons, and the ratio between the coefficients depends upon the
material in the individual voxel. If the x-ray CT information is to be used for correction of the PET emission information, the linear attenuation coefficient for each voxel
for 511-keV photons must be estimated from the linear attenuation coefficient (or,
equivalently, the CT number) for that voxel for x-rays from the CT system.
The linear attenuation coefficient is the product of the density of a material and
its mass attenuation coefficient. Table 19-4 lists the mass attenuation coefficients for
70-keV x-rays and 511-keV annihilation photons for several materials. The mass
attenuation coefficients in the CT energy range vary greatly with the atomic number
(Z) of the material because of the photoelectric effect. Although the interactions of
the x-rays from an x-ray CT system with soft tissue and bodily fluids are mainly by
Compton scattering, their interactions with higher atomic number materials such
as bone mineral, x-ray contrast material, and metal objects in the body, are by both
the photoelectric effect and Compton scattering. However, 511-keV annihilation
photons almost entirely interact with all these materials by Compton scattering. As
discussed in Chapter 3, the Compton scattering component of the linear attenuation
coefficient for photons of a specific energy is determined by the number of electrons
per volume, which in turn is largely determined by the density of the material.
The commonly used methods for estimating the linear attenuation coefficients
for 511-keV photons from the CT numbers (defined in Chapter 10) are based upon
making assumptions about the composition of the material in individual voxels using
738
TABLE 19-4COMPARISON OF MASS ATTENUATION COEFFICIENTS
FOR SOME TISSUES AND MATERIALS

MASS ATTENUATION COEFFICIENTS (cm2/g)
PHOTON ENERGY (keV)
SKELETAL MUSCLE
CORTICAL BONE
TITANIUM
IODINE
70
0.192
0.263
0.545
5.02
511
0.0951
0.0894
0.0811
0.0952
Hubbell JH, Seltzer SM. Tables of X-Ray Mass Attenuation Coefficients and Mass Energy-Absorption Coefficients 1 keV to 20 MeV for Elements Z=1 to 92 and 48 Additional Substances of Dosimetric Interest, NISTIR
5632, US Department of Commerce, May 1995.
For x-rays of a typical energy (70 keV) produced by a CT scanner, the likelihood of the photoelectric effect has
a large effect on mass attenuation coefficients and so they vary greatly with the atomic numbers of the material. However, for 511-keV annihilation photons, Compton scattering is by far the most likely interaction in
these materials and the mass attenuation coefficient varies little with atomic number. Note that attenuation
is determined by the linear attenuation coefficient, which is the product of the mass attenuation coefficient
and the density of the material.
their CT numbers. One method for estimating the linear attenuation coefficients for
511-keV photons from the CT numbers is to divide the voxels into two groups: those
with CT numbers less than a value such as 50 and those with CT numbers greater
than this value (Carney et al., 2006). Voxels with CT numbers less than this value are
assumed to contain only soft tissue, body fluids, gas, or a mixture of them, whereas
voxels with CT numbers greater than this value are assumed to contain bone mineral
and soft tissue. For each of these two groups, the linear attenuation coefficient for
511-keV photons is calculated from a linear equation relating these linear attenuation
coefficients to the CT number. For voxels in the low CT number group:
m511keV (9.6 10- 5 cm-1 ) (CTnumber 1,000)
This equation will yield the correct linear attenuation coefficients for air (CT
number 1000, 511 keV 0) and water (CT number 0, 511 keV 0.096 cm-1)
and approximately correct linear attenuation coefficients for most soft tissues. For
voxels in the high CT number group, the linear attenuation coefficient for 511-keV
photons is calculated from
m511keV = m (CTnumber ) + b
where m and b are empirically determined constants that differ with the kV used by
the x-ray CT system. In effect, the linear equation for the lower CT number voxels
uses the CT number to determine the density of the soft tissue in a voxel, whereas
the linear equation for the high CT number voxels uses the CT number to determine
the ratio of mass of bone mineral to mass of soft tissue in a voxel. These methods for
estimating the linear attenuation coefficients for 511-keV photons from CT numbers
are successful when the assumptions about the composition of material in a voxel are
valid, but can fail badly when these assumptions are not correct.
Incorrect estimation of the linear attenuation coefficients for 511-keV photons
can cause artifacts in the attenuation-corrected images. This commonly occurs when
there is a material, such as metal or concentrated x-ray contrast material, that does
not conform to the assumptions implicit in the method used to estimate attenuation
coefficients for 511-keV photons from the x-ray CT data. If the attenuation coefficient
for 511-keV photons is significantly overestimated, the artifact typically appears as a
falsely elevated radionuclide concentration. Such artifacts are discussed below.
739
Artifacts in PET/CT Imaging

Most things that cause artifacts in CT imaging, such as implanted metal objects that
cause star artifacts and patient motion, can in turn cause artifacts in attenuationcorrected PET/CT images. However, there are also artifacts, discussed below, that are
not artifacts in the PET images or the CT images themselves, but are caused by the
interactions of the two image sets.
Spatial Misregistration Artifacts

Despite the fact that the patient lies in the same position for both the PET and x-ray
CT scans, spatial misregistration of image information can still occur because the
PET and x-ray CT image information are not acquired simultaneously. Misregistration most commonly occurs due to organ motion, particularly in or near the thorax
due to respiration. The x-ray CT imaging is commonly performed during a single
breath-hold to produce images without the effects of respiratory motion, whereas the
PET imaging occurs during normal resting respiration. In particular, if the CT imaging is performed with breath-holding after full inspiration and the PET imaging is
acquired during shallow breathing, F-18 FDG avid lesions located in the liver close
to the diaphragm or in the chest wall may appear to be in the lungs in the fused PET/
CT images. Figure 19-28 shows a case of misregistration due to respiratory motion.
A partial solution to this problem is to acquire low-dose x-ray CT information
for image coregistration and attenuation correction during normal resting breathing
and, if needed, to acquire separate high-dose diagnostic quality CT images during
breath holding at maximal inspiration. Another partial solution is to acquire the CT
images with breath holding at midinspiration or midexpiration. Some systems allow
respiratory gating to be performed during the PET acquisition. Respiratory gating
can reduce respiratory motion artifacts and permit more accurate determination of
tissue volumes and activity concentrations, but increases the acquisition time.
Figure 19- 28 PET/CT study showing diaphragmatic misregistration with superimposition of a hypermetabolic liver nodule over the right lung base on the fused PET-CT. Left to right: coronal CT, attenuation-corrected PET, fused attenuation corrected PET-CT, and attenuation-corrected PET maximum intensity projection
images. CT shows no nodule in the right lung base, and attenuation-corrected PET and fused PET-CT images
show a clear band above the diaphragm due to misregistration. (Images and interpretation courtesy of George
M. Segall , M.D., VA Palo Alto Health Care System.)
740
Attenuation Correction Artifacts

When x-ray CT information is used for attenuation correction of the PET information, incorrect estimation of linear attenuation coefficients for 511-keV photons can
cause artifacts. As mentioned above under Attenuation Correction in PET/CT, material in the patient that does not meet the assumptions inherent in the method to
estimate attenuation coefficients for 511-keV photons from x-ray CT data can cause
significant errors in attenuation correction. If the attenuation coefficient for 511keV photons is significantly overestimated, the artifact typically appears as a falsely
elevated radionuclide concentration. This can occur where there are metallic objects
in the body, such as pacemakers, orthopedic devices, or body piercings, or concentrations of x-ray contrast material. Such an artifact is shown in Figure 19-29.
Voluminous metal objects, such as hip implants, in the patient may not cause such
attenuation artifacts displaying falsely elevated radionuclide concentration. Although
the attenuation coefficient for 511-keV photons is greatly overestimated, there is no
positron-emitting radionuclide in the metal object and so the attenuation correction
and image reconstruction process has no signal to incorrectly overamplify.
As mentioned above, x-ray contrast material, commonly containing iodine (intravascular contrast material) or barium (contrast material for GI studies) can also cause
errors in attenuation correction (Mawlawi et al., 2006; Otero et al., 2009). It appears
that such material, unless concentrated in a particular location, is not likely to have
clinically significant effects.
A focus of apparent enhanced uptake in the attenuation-corrected images can be
confirmed or refuted by a review of the corresponding uncorrected images. Spatial
misregistrations of CT and PET data can also cause errors in attenuation correction
(Goerres et al., 2003). Such an artifact appears at the level of the diaphragm (band of
falsely-low pixel values) in the middle two images in Figure 19-28.
FIGURE 19-29 Attenuation correction artifact caused by overestimation of attenuation by a metal object,
in this case a body piercing. CT image (A) shows a streaking artifact at the location of the metal object on the
left breast. Overestimation of attenuation causes the attenuation-corrected PET image (B) to display a falsely
elevated concentration of F-18 FDG at the location of the metal object. Notice that that concentration of F-18
FDG is not elevated in the PET image (C) that is not attenuation corrected. (Images courtesy of Sureshbabu W,
Mawlawi O. PET/CT imaging artifacts. J Nucl Med Technol 2005;33:156161.)
741
Truncation Artifacts
The radial FOVs of the CT and PET systems may differ significantly. Commonly,
the PET system has a much larger radial FOV than the CT, although at least one
manufacturer offers a system in which the PET and CT systems have equal FOVs.
If part of a patient, for example a patient of large girth or a patient imaged with
the arms by his or her side, extends outside the FOV of the CT, there will be error
in the attenuation correction of the PET images because attenuation by the portion of
the patient outside the CTs FOV is not considered.
SPECT/X-ray CT Systems
Several manufacturers provide imaging systems incorporating two scintillation camera heads capable of planar imaging and SPECT and an x-ray CT system with a single
patient bed. Some of these systems have very simple x-ray CT systems that provide
x-ray CT information for attenuation correction of the SPECT information and image
co-registration only, whereas others can produce diagnostic quality CT images.
The advantages and disadvantages of using the x-ray CT system for attenuation
correction and image coregistration are the same as in the case of the PET/CT systems
described above. The x-ray CT system acquires the attenuation information much more
quickly than does a system using radioactive sealed sources and the attenuation information has less statistical noise, but the linear attenuation coefficients for the energies
of the gamma rays must be estimated from the CT attenuation information. The methods for doing this are similar to those used in PET/CT, which were discussed earlier in
this chapter. Artifacts can occur when the calculation produces the wrong attenuation
coefficient. This can be caused by high atomic number material in the patient, such
as metal objects and concentrated contrast material. Furthermore, as in PET/CT, the
SPECT and x-ray CT information are not acquired simultaneously and patient organ
motion can result in the misregistrations of SPECT and PET image information.
A common use of SPECT/CT is myocardial perfusion imaging. In myocardial
perfusion imaging, nonuniform attenuation, particularly by the diaphragm and,
in women, the breasts, can cause apparent perfusion defects in the SPECT images.
Attenuation correction using the CT information has been reported to improve diagnostic accuracy by compensating for these artifacts. However, spatial misregistration
between the heart in the CT and SPECT images can cause artifacts (Goetze et al.,
2007). An important quality assurance step in myocardial perfusion SPECT/CT is to
verify alignment of the SPECT and CT image information.
PET/MRI Systems
Dual modality PET/MRI systems have very recently been developed. These include
small systems for imaging of animals and larger systems for imaging humans for
research. As in PET/CT, the MRI system provides tomographic images with excellent
depiction of anatomic detail and the PET information is spatially coregistered with
the MRI information. The MRI information is also used to estimate attenuation coefficients in each voxel of the patient for correction of the PET information for attenuation of the annihilation photons.
A problem that must be addressed in the design of PET/MRI imaging systems is
that photomultiplier tubes (PMTs) are adversely affected by magnetic fields, particularly very strong ones such as those produced by MRI magnets. Furthermore, PMTs
are very bulky. Two methods have been used to address this problem. In one approach,
the rings of PET detectors are inside the bore of the MRI systems main magnet. These
742
detectors consist of scintillation crystals optically coupled to avalanche photodiodes,

which are not affected by the strong magnetic field. (Avalanche photodiodes were
briefly discussed in Chapter 17.) In the other approach, the PET system is at one end
of the patient couch and the MRI system is at the other end of the patient couch, with
the two systems being separated by a distance of about 2 m. In this latter approach,
the PET detectors consist of scintillation crystals coupled to PMTs. A system with the
PET detectors inside the MRI permits simultaneous image acquisition, whereas the
other design requires sequential imaging.
Quality Control of Dual Modality Imaging Systems

There are three aspects to the quality control of dual modality imaging systems.
There is quality control testing of the SPECT or PET imaging system; these were
discussed in the relevant sections earlier in this chapter. Then, there should be the
quality control testing of the CT or MRI system. Lastly, there should be testing of
the operation of the combined systems. The two major items to be tested are the
accuracy of the spatial coregistration of the two modalities and the accuracy of the
attenuation correction.
19.5 Clinical Aspects, Comparison of PET and SPECT,
and Dose
As mentioned in the first chapter of this textbook, nuclear medicine imaging produces images depicting function. The function depicted may be a mechanical function, such as the ability of the heart to contract, but very often is a physiological or
biochemical function. However, a disadvantage is that the images may lack sufficient
anatomic information to permit determination of the organ or tissue containing a
feature of interest. Fusion of the nuclear medicine images with those from another
modality, such as CT, providing good depiction of anatomy can resolve this problem.
Another disadvantage of imaging radiopharmaceuticals is that attenuation by the
patient of the emitted photons can cause artifacts. CT attenuation information can be
used to largely correct these artifacts.
F-18 Fluorodeoxyglucose and Other PET

Radiopharmaceuticals
As discussed in the previous chapter, the utility of nuclear medicine is as much determined by the radiopharmaceutical as it is by the imaging device. For example, the
success of the Anger scintillation camera was in part due to the development of many
radiopharmaceuticals incorporating Tc-99m.
An advantage of PET is that there are positron-emitting radioisotopes of the
common biochemical elements carbon, nitrogen, and oxygen, but these have short
half-lives (20, 10, and 2 min, respectively). The use of radiopharmaceuticals incorporating them requires a nearby or on-site cyclotron and so their use has largely been
restricted to research.
The widespread clinical adoption of PET and now PET/CT has been driven
by the remarkable radiochemical (F-18)2-fluoro-2-deoxy-d-glucose (F-18 FDG).
F-18 FDG is a glucose analog; it is transported by facilitated diffusion into cells
that utilize glucose and phosphorylated, as is glucose. However, the phosphorylated F-18 FDG cannot proceed further along the metabolic path of glucose and
743
therefore remains in the cells. Thus, F-18 accumulates in cells at a rate proportional to local extracellular concentration and glucose uptake. (Some cells have an
enzyme that dephosphorylates glucose-6-phosphate and some F-18 FDG escapes
these cells.) F-18 has a half-life of 110 min, permitting a cyclotron-equipped
radiopharmacy to provide F-18 FDG throughout a metropolitan area or even over
larger distances. FDG has shown clinical utility for differentiating Alzheimers
disease from other forms of dementia, identifying foci responsible for epileptic seizures, and for differentiating hypoperfused myocardium from scar tissue.
However, its most common application today is in oncology, in which it is used
to evaluate and assess the efficacy of treatment for the many forms of cancer that
accumulate glucose. However, F-18 FDG is only of limited use in some forms of
cancer, notably prostate cancer.
The FDA has approved a few other PET radiopharmaceuticals, which are
described in Chapter 16. Other PET radiopharmaceuticals are under study.
Radiation Doses
As mentioned in Chapter 16, the radiation doses from the administration of a radiopharmaceutical depends on the physiological behavior of the radiopharmaceutical,
the physical half-life of the radionuclide, the activity administered, and the types and
energies of the radiations emitted. Positrons deposit nearly all of their kinetic energies within millimeters of the sites of their emission. Much, but not all, of the energy
of the annihilation photons escapes the patient. To keep the doses to patients sufficiently low, the radionuclides commonly used in PET have been restricted to ones
with relatively short half-lives. The intravenous administration of 370 MBq (10 mCi)
of F-18 FDG to an adult imparts an effective dose of about 7 mSv (Appendix F-2).
The radiation dose from a PET/CT or SPECT/CT examination includes that of the
CT scan. In oncologic PET/CT, the CT scan is typically performed on a length of the
body from just below the eyes to the symphysis pubis. The effective dose from this
CT scan is relatively large because of the length of the body that is scanned. The CT
scan can be a high x-ray tube current scan that produces images of diagnostic quality, or it can be a low dose scan that produces images only for attenuation correction
and anatomic correlation. The effective dose from the former may be about 16 mSv,
whereas that from the latter may be about 4 mSv (Brix et al., 2005). Innovations in
CT such as automatic tube current modulation and iterative image reconstruction
may reduce these doses.
Comparison of SPECT and PET

In single photon emission imaging, the spatial resolution and the detection efficiency
are primarily determined by the collimator. Both are ultimately limited by the compromise between collimator efficiency and collimator spatial resolution that is a consequence of collimated image formation. It is the use of ACD instead of collimation
that makes the PET scanner much more efficient than the scintillation camera and
also yields its superior spatial resolution.
In systems that use collimation to form images, the spatial resolution rapidly
deteriorates with distance from the face of the imaging device. This causes the spatial
resolution to deteriorate from the edge to the center in transverse SPECT images. In
contrast, PET is not subject to this limitation and the spatial resolution in a transverse
PET image is best in the center.
744
TABLE 19-5 COMPARISON OF SPECT AND PET

SPECT
PET
Principle of projection Collimation.

data collection
Annihilation coincidence
detection (ACD).
Transverse image
reconstruction
Iterative methods or filtered

backprojection.
Iterative methods or filtered

backprojection.
Radionuclides
Any emitting x rays, gamma rays,

or annihilation photons. Optimal
performance for photon energies
of 100200 keV.
Positron emitters only.
Spatial resolution
Depends upon collimator and

camera orbit.
Relatively constant across

transaxial image, best at center.
Within a transaxial image, the

resolution in the radial direction is
relatively uniform, but the
tangential resolution is degraded
toward the center.
Typically 4.55 mm FWHM at

center.
Typically about 10 mm FWHM at

center for a 30 cm diameter orbit
and 99mTc.
Larger camera orbits produce
worse resolution.
Attenuation
Attenuation less severe.

Radioactive attenuation correction
sources or x-ray CT can correct for
attenuation.
Attenuation more severe.

Radioactive attenuation
correction sources or x-ray CT can
correct for attenuation.
The cost of a dual-head SPECT system is typically a few hundred thousand dollars. The cost of a SPECT-CT system depends greatly on the capabilities of the CT
system. As mentioned previously, some have slow low-dose CT systems that are not
capable of producing diagnostic quality CT images, whereas others have multirow
detector, subsecond rotation CT systems that can produce diagnostic quality images.
The cost of the latter system may be more than twice that of a SPECT system without
CT. The cost of a PET/CT system is about twice that of a SPECT/CT system capable
of diagnostic quality CT imaging. Table 19-5 compares SPECT and PET systems.
Quantitative Imaging
It is desirable, for research studies and many clinical studies, that each pixel value
in a SPECT or PET image be proportional to the number of nuclear transformations
occurring during imaging in the corresponding voxel of the patient. For example,
in myocardial perfusion imaging, it is desirable that the count in each pixel depicting left ventricular myocardium be proportional to the concentration of radioactive
tracer in the corresponding voxel in the patients heart.
The primary factors limiting the quantitative accuracy of SPECT and PET are
imperfect spatial resolution and photon attenuation and scattering in the patient.
At high interaction rates in the imaging system, other effects such as dead-time
count losses and, in PET, random coincidences may also contribute to divergence
from proportionality. Yet another reason is patient motion. Because SPECT and PET
imaging occur for lengthy time intervals, spatial resolution is degraded by motion,
745
articularly respiratory motion. PET with radioactive sources for attenuation corp
rection and PET/CT are considered quantitative imaging modalities and the reconstructed voxel values can be calibrated in units of activity.
The partial volume effect is a reduction in the apparent activity concentration in
smaller objects caused by averaging of the actual activity concentration with that of
surrounding tissues. Causes of the partial volume effect are blurring caused by the
imaging device; blurring by the tomographic reconstruction process (e.g., the spatial
filtering used in filtered backprojection or after iterative reconstruction); and the
sizes of the pixels in the resultant images. PETs higher spatial resolution causes it to
suffer less from the partial volume effect than SPECT.
A physiological model was developed by Sokoloff and colleagues that enables the
rate of local tissue glucose utilization to be calculated from the amount of radiolabeled FDG that accumulates in the tissue (Sokoloff et al., 1977). The model has been
extended to permit non-invasive measurements of this rate using F-18 FDG and
PET imaging (Phelps et al., 1979). This latter method has been used extensively in
research studies of the brain, heart, and other organs. However, the method requires
blood sampling, which has hindered its clinical use.
The standardized uptake value (SUV) is a calculated quantity used in F-18 FDG
imaging in an attempt to normalize the uptake of F-18 FDG for administered activity, radioactive decay from the time of dosage administration, and patient body mass
(Keyes, 1995; Thie, 2004). Some physicians use the SUV to assist in characterizing tissues that accumulate the radiopharmaceutical, particularly in distinguishing
benign from malignant lesions or in monitoring tumor therapy. It is defined as
SUV
activity concentration in a voxel or group of voxels

activity administered/body mass
where either the administered activity or the activity concentration is corrected for
radioactive decay from the time of dosage assay until imaging.
The SUV has units of density, for example, g/cm3. Some institutions use lean
body mass or body surface area instead of body mass in the denominator of this
equation. Because the SUV is not based upon a physiological model, such as the
Sokoloff model, it is regarded as an approximate or semiquantitative index of FDG
uptake.
The SUV measurement is affected by many factors:
1. Accuracy of the assay of activity to be administered. This includes correction for
activity remaining in the syringe after administration
2. Extravasation of activity during administration, which occurs in a small fraction
of administrations
3. Calibration of the PET/CT and accuracy of attenuation correction
4. Elapsed time from activity administration until imaging, including correct
recording of time of administration
5. Accuracy of patient body mass
6. Physiological state of patient at time of administration. This includes preparation of the patient, including fasting before administration and possibly insulin
administration
7. Patient body composition (e.g., ratio of fatty to lean tissue)
8. Size of lesion. The partial volume affect can reduce the measured SUV for very
small lesions
9. Motion, such as respiratory motion, during image acquisition, particularly for
smaller lesions
10. User region-of-interest (ROI) selection
746
Clinicians using the SUV for making decisions regarding patient workup, staging,
treatment, and treatment planning should discuss with their nuclear medicine physicians the potential uncertainties in the SUV as it is determined in their nuclear
medicine service or PET/CT provider and apply caution in using SUVs from different
institutions.
SUGGESTED READINGS
Bax JJ, Wijns W. Editorialfluorodeoxyglucose imaging to assess myocardial viability: PET, SPECT, or
gamma camera coincidence imaging. J Nucl Med 1999;40:18931895.
Carney JPJ, Townsend DW, Rappoport V, Bendriem B. Method for transforming CT images for attenuation
correction in PET/CT imaging. Med Phys 2006;33(4):976983.
Cherry SR, Phelps ME. Positron emission tomography: methods and instrumentation. In: Sandler MP,
et al., eds. Diagnostic nuclear medicine. 4th ed. Baltimore, MD: Lippincott Williams & Wilkins, 2003:6183.
EANM Physics Committee: Sokole EB, Pachcnska A, Britten A with contribution from the EANM Working
Group on Nuclear Medicine Instrumentation Quality Control: Georgosopoulou ML, Tindale W, Klett R.
Routine quality control recommendations for nuclear medicine instrumentation Eur J Nucl Med Mol
Imaging 2010;37:662671.
Gelfand MJ, Thomas SR. Effective use of computers in nuclear medicine. New York, NY: McGraw-Hill, 1988.
Groch MW, Erwin WD. SPECT in the Year 2000: Basic principles. J Nucl Med Technol 2000;28:233244.
Groch MW, Erwin WD, Bieszk JA. Single photon emission computed tomography. In: Treves ST, ed.
Pediatric nuclear medicine. 2nd ed. New York, NY: Springer-Verlag, 1995:3387.
Hines H, Kayayan R, Colsher J, Hashimoto D, Schubert R, Fernando J, Simcic V, Vernon P, Sinclair RL.
NEMA recommendations for implementing SPECT instrumentation quality control. J Nucl Med
2000;41:383389.
Patton JA, Turkington TG. SPECT/CT physical principles and attenuation correction. J Nucl Med Technol
2008;36:110.
Sureshbabu W, Mawlawi O. PET/CT imaging artifacts. J Nucl Med Technol 2005;33:156161.
Thie JA. Understanding the standardized uptake value, its methods, and implications for usage. J Nucl Med
2004;45(9):14311434.
Yester MV, Graham LS, eds. Advances in nuclear medicine: the medical physicists perspective. Proceedings
of the 1998 Nuclear Medicine Mini Summer School, American Association of Physicists in Medicine,
June 2123, 1998, Madison, WI.
Zaidi H, ed. Quantitative analysis in nuclear medicine imaging. New York, NY: Springer, 2006.
Zanzonico P. Routine quality control of clinical nuclear medicine instrumentation: a brief review. J Nucl
Med 2008;49:11141131.
SELECTED REFERENCES
Boren EL Jr., Delbeke D, Patton JA, Sandler MP. Comparison of FDG PET and positron coincidence
detection imaging using a dual-head gamma camera with 5/8-inch NaI(Tl) crystals in patients with
suspected body malignancies. Eur J Nucl Med Mol I 1999;26(4):379387.
Brix G, Lechel U, Glatting G, Ziegler SI, Mnzing W, Mller SP, Beyer T. Radiation exposure of patients
undergoing whole-body dual-modality 18F-FDG PET/CT examinations. J Nucl Med 2005;46:608613.
Chang LT. A method for attenuation correction in radionuclide computed tomography. IEEE Trans Nucl
Sci 1978;NS-25:638.
Delbeke D, Patton JA, Martin WH, Sandler MP. FDG PET and dual-head gamma camera positron coincidence
detection imaging of suspected malignancies and brain disorders. J Nucl Med 1999;40(1):110117.
Goetze S, Brown TL, Lavely WC, Zhang Z, Bengel FM. Attenuation correction in myocardial perfusion
SPECT/CT: effects of misregistration and value of reregistration. J Nucl Med 2007;48(7):10901095.
Hendel RC, et al. The value and practice of attenuation correction for myocardial perfusion SPECT
imaging: a joint position statement from the American Society of Nuclear Cardiology and the Society
of Nuclear Medicine. J Nuclear Med 2002;43(2):273280.
Hudson HM, Larkin RS. Accelerated image reconstruction using ordered subsets of projection data. IEEE
Trans Med Imaging 1994;13:601609.
Keyes JW Jr. SUV: Standard Uptake or Silly Useless Value? J Nucl Med 1995; 36:18361839.
Kuhl DE, Edwards RQ. Image separation radioisotope scanning. Radiology 1963;80:653662.
Liu Y-H, Lam PT, Sinusas AJ, Wackers FJTh. Differential effect of 180 and 360 acquisition orbits on the
accuracy of SPECT imaging: quantitative evaluation in phantoms. J Nuclear Med 2002;43(8):11151124.
747
Macfarlane DJ, Cotton L, Ackermann RJ, Minn H, Ficaro EP, Shreve PD, Wahl RL. Triple-head SPECT with
2-[fluorine-18]fluoro-2-deoxy-D-glucose (FDG): initial evaluation in oncology and comparison with
FDG PET. Radiology 1995;194:425429.
Masood Y, et al. Clinical validation of SPECT attenuation correction using x-ray computed tomography
derived attenuation maps: multicenter clinical trial with angiographic correlation. J Nuclear Cardiol
2005;12(6);676686.
Mawlawi O, Erasmus JJ, et al. Quantifying the effect of iv contrast media on integrated PET/CT: clinical
evaluation. AJR 2006;186:308319.
National Electrical Manufacturers Association, NEMA NU 1-2007, Performance Measurements of Gamma
Cameras.
National Electrical Manufacturers Association, NEMA NU 2-2007, Performance Measurements of Positron
Emission Tomographs.
Nichols KJ. Editorial: How serious a problem for myocardial perfusion assessment is moderate
misregistration between SPECT and CT? J Nuclear Cardiol 2007;14(2):150152.
Otero HJ, Yap JT, et al. Evaluation of low-density neutral oral contrast material in PET/CT for tumor
imaging: results of a randomized clinical trial. AJR 2009;193:326332.
Oturai PS, Mortensen J, Enevoldsen H, Eigtved A, Backer V, Olesen KP, Nielsen HW, Hansen H, Stentoft P,
Friberg L. -camera 18F-FDG PET in diagnosis and staging of patients presenting with suspected lung
cancer and comparison with dedicated PET. J Nucl Med 2004;45(8):13511357.
Phelps ME, Huang SC, Hoffman EJ, Selin C, Sokoloff L, Kuhl DE. Tomographic measurement of local
cerebral glucose metabolic rate in humans with (F-18)2-fluoro-2-deoxy-D-glucose: validation of
method. Ann Neurol 1979;6:371388.
Sokoloff L, Reivich M, Kennedy C, Des Rosiers MH, Patlak CS, Pettigrew KD, Sakurada O, Shinohara M. The
[14C]deoxyglucose method for the measurement of local cerebral glucose utilization: theory, procedure,
and normal values in the conscious and anesthetized albino rat. J Neurochem 1977;28:897916.
Tiepolt C, Beuthien-Baumann B, Hlises R, Bredow J, Khne A, Kropp J, Burchert W, Franke WG. 18F-FDG
for the staging of patients with differentiated thyroid cancer: comparison of a dual-head coincidence
gamma camera with dedicated PET. Ann Nucl Med 2000;14(5):339345.
Yutani K, Tatsumi M, Shiba E, Kusuoka H, Nishimura T. Comparison of dual-head coincidence gamma
camera FDG imaging with FDG PET in detection of breast cancer and axillary lymph node metastasis.
J Nucl Med 1999;40(6):10031008.
Zhang H, Tian M, Oriuchi N, Higuchi T, Tanada S, Endo K. Oncological diagnosis using positron
coincidence gamma camera with fluorodeoxyglucose in comparison with dedicated PET. Brit J Radiol
2002;75:409416.
Zimny M, Kaiser HJ, Cremerius U, Sabri O, Schreckenberger M, Reinartz P, Bll U. F-18-FDG positron
imaging in oncological patients: gamma camera coincidence detection versus dedicated PET. Nuklearmed
1999;38(4):108114.
Section
IV
Radiation Biology
and Protection
Chapter
20
Radiation Biology
20.1 Overview
Rarely have beneficial applications and hazards to human health followed a major
scientific discovery more rapidly than with the discovery of ionizing radiation.
Shortly after Roentgens discovery of x-rays in 1895 and Becquerels discovery of
natural radioactivity in 1896, adverse biological effects from ionizing radiation were
observed. Within months after their discovery, x-rays were being used in medical
diagnosis and treatment. Unfortunately, the development and implementation of
radiation protection techniques lagged behind the rapidly increasing use of radiation
sources. Within the first 6 months of their use, several cases of erythema, dermatitis, and alopecia were reported among x-ray operators and their patients. Becquerel
himself observed radiation induced erythema on his abdomen from a vial of radium
he carried in his vest pocket during a trip to London to present his discovery. Many
years later, this effect was referred to as a Becquerel burn. The first report of a skin
cancer ascribed to x-rays was in 1902, to be followed 8 years later by experimental confirmation. However, it was not until 1915 that the first radiation protection
recommendations were made by the British Roentgen Society, followed by similar
recommendations from the American Roentgen Ray Society in 1922.
The study of the action of ionizing radiation on healthy and diseased tissue is the
scientific discipline known as radiation biology. Radiation biologists seek to understand
the nature and sequence of events that occur following the absorption of energy from
ionizing radiation, the biological consequences of any damage that results, and the
mechanisms that enhance, compensate for, or repair the damage. A century of radiobiologic research has amassed more information about the effects of ionizing radiation
on living systems than is known about almost any other physical or chemical agent.
This chapter reviews the consequences of ionizing radiation exposure, beginning
with the chemical basis by which radiation damage is initiated and its subsequent
effects on cells, organ systems, and the whole body. This is followed by a review of
the concepts and risks associated with radiation-induced carcinogenesis, hereditary
effects, and the special concerns regarding radiation exposure of the fetus.
Determinants of the Biologic Effects of Radiation

There are many factors that determine the biologic response to radiation exposure. In
general, these factors include variables associated with the radiation source and the
biologic system being irradiated. The identification of these biologic effects depends on
the method of observation and the time following irradiation. Radiation-related factors
include the absorbed dose (quantity) and dose rate as well as the type and energy (quality) of the radiation. The radiosensitivity of a complex biologic system is determined by
a number of variables some of which are inherent to the cells themselves while others
relate to the conditions in the cells at the time of irradiation. Damage observed at the
molecular or cellular level may or may not result in clinically detectable adverse effects.
751
752
Section IV Radiation Biology and Protection
urthermore, although some responses to radiation exposure appear instantaneously

F
or within minutes to hours, others take weeks, years, or even decades to appear.
Classification of Biologic Effects

Biologic effects of radiation exposure can be classified as either stochastic or deterministic. A stochastic effect is one in which the probability of the effect occurring, (rather
than its severity), increases with dose. Radiation-induced cancer and hereditary
effects are stochastic in nature. For example, the probability of radiation-induced
leukemia is substantially greater after an exposure to 1 Gy than to 10 mGy, but there
will be no difference in the severity of the disease if it occurs. Stochastic effects are
believed not to have a dose threshold, because damage to a few cells or even a single
cell could theoretically result in production of the disease. Therefore even minor
exposures may carry some, albeit small, increased risk (i.e., increased probability of
radiation induced cancer or genetic effect). It is this basic, but unproven, assumption
that risk increases with dose and there is no threshold dose below which the magnitude of the
risk goes to zero, that is the basis of modern radiation protection programs, the goal of
which is to keep exposures as low as reasonably achievable (see Chapter 21). Stochastic effects are regarded as the principal health risk from low-dose radiation, including
exposures of patients and staff to radiation from diagnostic imaging procedures.
If a radiation exposure is very high, the predominant biologic effect is cell killing,
which presents clinically as degenerative changes in the exposed tissue. In this case, the
effects are classified as deterministic for which the severity of the injury, (rather than its
probability of occurrence), increases with dose. Deterministic effects (also referred to as
tissue reactions) differ from stochastic effects in that they require much higher doses to
produce a clinically observable effect and there is a threshold dose below which the effect
does not occur or is subclinical. Skin erythema, fibrosis, and hematopoietic damage
are some of the deterministic effects that can result from large radiation exposures. As
there is substantial individual variability in response to radiation, the threshold dose
is really just an approximation of the dose that would likely result in the specified effect.
Many of these effects are discussed in the sections entitled Response of Organ Systems
to Radiation and The Acute Radiation Syndrome. Deterministic effects can be caused
by severe radiation accidents and can be observed in healthy tissue that is unavoidably irradiated during radiation therapy. However, with the exception of some lengthy,
fluoroscopically guided interventional procedures (see Chapter 9), they are unlikely to
occur as a result of routine diagnostic imaging procedures or occupational exposure.
20.2 Interaction of Radiation with Tissue

As discussed in Chapter 3, x-ray and gamma-ray photon interactions in tissue, as well
as radiations emitted during radionuclide decay, result in the production of energetic
electrons. These electrons transfer their kinetic energy to their environment via excitation, ionization, and thermal heating. Energy is deposited randomly and rapidly
(in less than 108 seconds) and the secondary ionizations set many more low-energy
electrons in motion causing additional excitation and ionization along the path of
the initial energetic electron. For example, a single 30 keV electron, set in motion
following the photoelectric absorption of a single x-ray or gamma-ray photon, can
result in the production of over 1,000 low-energy secondary electrons (referred to as
delta rays), each of which may cause additional excitation or ionization events in the
tissue. This chain of ionizations ultimately gives rise to subexcitation electrons (i.e.,
electrons with kinetic energies less than the first excitation potential of liquid water,
753
Chapter 20 Radiation Biology
7.4 eV) that become thermalized as they transfer their remaining kinetic energy by
vibrational, rotational, and collisional energy exchanges with the water molecules.
Observable effects such as chromosome breakage, cell death, oncogenic transformation, and acute radiation sickness, all have their origin in radiation-induced chemical
changes in important biomolecules.
Low Energy Electrons and Complex Damage

The delta rays and other lower energy electrons, set in motion following an initial
ionizing event, result in a unique ionization pattern in which closely spaced ionizations occur over a very short range (~4 to 12 nm) along the path of the primary
ionization track. The energy deposition (~100 ev) along the shorter tracks referred to
as spurs, whose diameters are approximately 4 to 5 nm, result in an average of three
ionizing events. It is estimated that 95% of the energy deposition events from x-rays
and gamma rays occurs in spurs (Hall, 2006). Longer and less frequent pear shaped
tracks called blobs deposit more energy (~300 to 500 ev) and thus on average result in
more ionization events (~12 ion pairs) over their path (~12 nm). The high concentrations of reactive chemical species (such as free radicalsdiscussed below) produced
by these spurs and blobs increase the probability of molecular damage at these locations (Fig. 20-1A,B). If ionizing events occur near the DNA, whose diameter (~2 nm)
Complex Cluster
DNA Damage
Low-LET tracks
in cell nucleus
for example, from
gamma rays
Low-energy Electron track
H
OH
OH
OH
OH
OH
H
300 eV
electron
A dose of 1 Gy
corresponds to
about 1000 tracks
= ionized molecule
= excited molecule
OH
B
~8,000 nm
~2 nm
FIGURE 20-1 A. Low-LET radiation like x-rays and gamma rays is considered sparsely ionizing on average;
however, a majority of the radiation energy is deposited in small regions (on the scale of nanometers) via denser
clusters of ionizations from low-energy secondary electrons. This illustration depicts primary and secondary electron tracks producing clusters of ionization events. The calculated number of tracks is based on a cell nucleus
with a diameter of 8 mm. The track size is enlarged relative to the nucleus to illustrate the theoretical track
structure. B. A segment of the electron track is illustrated utilizing a Monte Carlo simulation of clustered damage
produced by ionizations and excitations along the path of a low-energy (300 eV) electron. Excitation and ionization along with secondary electrons are shown until the electron energy drops below the ionization potential
of water (~ 10 eV). C. DNA double helix drawn on the same scale as the ionization track. Complex clustered
damage can result from closely spaced damage to the DNA sugar-phosphate backbone and bases from both
direct ionizations and diffusion of OH radicals produced by the radiolysis of water molecules in close proximity
(few mm) with the DNA (i.e., indirect effect). Multiple damaged sites are shown as green, or orange, explosion
symbols that denote DNA strand breaks, or damaged bases, respectively. In this example, the result is a complex double strand break, consisting of three strand breaks and three damaged bases, all within ten base pairs
along the DNA. This type of complex DNA lesion is more difficult for the cell to repair and can lead to cell death,
impaired cell function, or transformations with oncogenic potential, (Adapted from Goodhead 2006, 2009).
754
is on the same order as that of these short ionization tracks, they can produce damage
in multiple locations in the DNA in close proximity to one another. These lesions, initially referred to as locally multiply damaged sites are more difficult for the cell to repair
or may be repaired incorrectly, Figure 20-1C (Goodhead, 1994; Ward, 1988).
Synonyms for this type of damage in more common use today include clustered
damage, complex damage, and multiply damaged sites (MDS). While endogenous processes
mainly produce isolated DNA lesions, the complex clustered damage, in which groups
of several damaged nucleotides occur within one or two helical turns of the DNA, is a
hallmark of ionizing radiationinduced DNA damage. While the ionization pattern and
radiation chemistry of the microenvironment produced by radiation is different from
other oxidative events in the cell of endogenous (e.g., oxidative metabolism) and exogenous (e.g., chemotherapy) origin, the functional changes produced in molecules, cells,
tissues, and organs cannot be distinguished from damage produced by other sources.
Repair of radiation damage occurs at molecular, cellular, and tissue levels. A complex
series of enzymes and cofactors repair most radiation-induced DNA lesions within hours
and, to the extent possible, damaged cells are often replaced within days following irradiation. However, clinical manifestation of radiation-induced damage may appear over a
period of time that varies from minutes to weeks and even years depending on the radiation dose, cell type, and the nature and scope of the damage. Only a fraction of the radiation energy deposited brings about chemical changes; the vast majority of the energy is
deposited as heat. The heat produced is of little biologic significance compared with the
heat generated by normal metabolic processes. For example, it would take more than
4,000 Gy, a supralethal dose, to raise the temperature of tissue by 1C.
Free Radical Formation and Interactions

Radiation interactions that produce biologic changes are classified as either direct or
indirect. The change is said to be due to direct action if a biologic macromolecule such
as DNA, RNA, or protein becomes ionized or excited by an ionizing particle or photon
passing through or near it. Indirect action refers to effects that are the result of radiation interactions within the medium (e.g., cytoplasm) that create mobile, chemically
reactive species that in turn interact with nearby macromolecules. Because approximately 70% of most cells in the body are composed of water, the majority (~60%) of
radiation-induced damage from medical irradiation is caused by radiation interactions
with water molecules. The physical and chemical events that occur when radiation
interacts with a water molecule lead to the formation of a number of different highly
reactive chemical species. Initially, water molecules are ionized to form, H2O and e.
The e rapidly thermalizes and becomes hydrated, with a sphere of water molecules
orienting around the e to form the hydrated or aqueous electron (eaq). The eaq then
reacts with another water molecule to form a negative water ion (H2O eaq H2O).
These waters ions very unstable; each rapidly forms another ion and a free radical:
H2O H2O H3O OH (Hydroxyl Radical)
H2O OH H(Hydrogen Radical)
Free radicals, denoted by a dot next to the chemical symbol, are atomic or molecular
species that have unpaired orbital electrons. Thus, free radicals can be radical ions
(e.g., H2O and H2O), or electrically neutral (OH). The hydrogen and hydroxyl
radicals can be created by other reaction pathways, the most important of which is the
radiation-induced excitation and disassociation of a water molecule (H2O* excitation
H and OH). The H and OH ions do not typically produce significant biologic
damage because of their extremely short lifetimes (~1010 seconds) and their tendency
755
to recombine to form water. Free radicals are extremely reactive chemical species that
can undergo a variety of chemical reactions. Free radicals can combine with other free
radicals to form nonreactive chemical species such as water (e.g., H OH H2O),
in which case no biologic damage occurs, or with each other to form other molecules
such as hydrogen peroxide (e.g., OH OH H2O2), which is highly toxic to the
cell. However, for low linear energy transfer (LET) radiation like x-rays and gamma
rays, the molecular yield of H2O2 is low and the majority of indirect effects are due to
the interactions of the hydroxyl radicals with biologically important molecules.
The damaging effect of free radicals is enhanced by the presence of oxygen. Oxygen stabilizes free radicals and reduces the probability of free radical recombination
into nontoxic chemicals such as water or molecular hydrogen. Oxygen can combine
with the hydrogen radical to form a highly reactive oxygen species (ROS) such as the
hydroperoxyl radical (e.g., H O2 HO2). Free radicals can act as strong oxidizing or reducing agents by combining directly with macromolecules. Free radicals can
attack biomolecules (R) in a number of ways including hydrogen abstraction (RH
OH R H2O) and OH addition (R OH ROH). Chemical repair of the
damaged biomolecules can occur via radical recombination (e.g., R H RH)
or more commonly, by hydrogen donation from thiol compounds (RSH R RH
RS), producing the much less reactive or damaging thiyl radical. In the presence
of oxygen, repair is inhibited by the chemical transformation into peroxyradicals
(R O2 RO2). Although their lifetimes are limited (less than 10-5 seconds), free
radicals can diffuse sufficiently far in the cell (on the order of ~4 nm), to produce
damage at locations other than their origin. Free radical-induced damage to DNA
is the primary cause of biologic damage from low-LET radiation. While radiation
exposure from medical imaging does result in some direct ionization of critical cellular targets, approximately two thirds of the total radiation damage is due to the free
radical-mediated indirect effects of ionizing radiation.
Many enzymatic repair mechanisms exist within cells that are capable, in most cases,
of returning the cell to its preirradiated state. For example, if a break occurs in a single
strand of DNA, the site of the damage is identified and the break may be repaired by
rejoining the broken ends. If the damage is too severe or these repair mechanisms are
compromised or overwhelmed by excessive radiation exposure, the DNA could suffer
mutations. The clinical consequence of such DNA damage depends on a number of variables. For example, if the damage were to the DNA at a location that prevented the cell
from producing albumin, the clinical consequences would be insignificant considering
the number of cells remaining with the ability to produce this serum protein. If, however,
the damage were to the DNA at a location that was responsible for controlling the rate of
cell division (e.g., in an oncogene or tumor suppressor gene), the clinical consequences
could be the formation of a tumor or cancer. Heavily irradiated cells, however, often die
during mitosis, thus preventing the propagation of seriously defective cells. Figure 20-2
summarizes the physical and biologic responses to ionizing radiation.
Experiments with cells and animals have shown that the biologic effect of radiation depends not only on factors such as the dose, dose rate, environmental conditions at the time of irradiation, and radiosensitivity of the biologic system, but also
on the spatial distribution of the energy deposition at the molecular level (microdosimetry). The LET is a parameter that describes the average energy deposition
per unit path length of the incident radiation (see Chapter 3). Although all ionizing
radiations are capable of producing the same types of biologic effects, the magnitude
of the effect per unit dose differs. To evaluate the effectiveness of different types and
energies of radiation and their associated LETs, experiments are performed that compare the dose required for the test radiation to produce the same specific biologic
756
Indirect Action
Direct Action
Repair
Radiation
Radiation
H2O
H+ OH H2O*
Ionization and excitation
H ; OH ; HO2 ; H2O; (e
aq); H2O2
DNA
Molecular Damage
Free Radicals and
Hydrogen Peroxide
Mutation
Genetic
Biological response
Somatic
Death
Teratogenic
1017 to 105 seconds
minutes to decades
FIGURE 20-2 Physical and biologic responses to ionizing radiation. Ionizing radiation causes damage either directly
by damaging the molecular target or indirectly by ionizing water, which in turn generates free radicals that attack
molecular targets. The physical steps that lead to energy deposition and free radical formation occur within 105 to
106 seconds, whereas the biologic expression of the physical damage may occur from seconds to decades later.
response produced by a particular dose of a reference radiation (typically, x-rays

produced by a potential of 250 kV). The term relating the effectiveness of the test
radiation to the reference radiation is called the relative biological effectiveness (RBE).
The RBE is defined, for identical exposure conditions, as follows:
RBE
Dose of 250-kV x-rays required to produce effect X

Dose of test radiation required to produce effect X
The RBE is initially proportional to LET: As the LET of the radiation increases, so does
the RBE (Fig. 20-3). The increase is attributed to the higher specific ionization (i.e., ionization density) associated with high-LET radiation (e.g., alpha particles) and its relative
advantage in producing cellular damage (increased number and complexity of clustered
DNA lesions) compared with low-LET radiation (e.g., x-rays and gamma rays). However, beyond approximately 100 keV/mm in tissue, the RBE decreases with increasing
LET, because of the overkill effect. Overkill (or wasted dose) refers to the deposition of
radiation energy in excess of that necessary to produce the maximum biologic effect.
The RBE ranges from less than 1 to more than 20. For a particular type of radiation, the
RBE depends on the biologic end point being studied. For example, chromosome aberrations, cataract formation, or acute lethality of test animals may be used as end points.
Compared to high-energy gamma rays, the increased effectiveness of diagnostic x-rays
in producing DNA damage is suggested not only by the differences in their microdosimetric energy deposition patterns but has also been demonstrated experimentally with
an RBE of about 1.5 to 3. However, these differences do not necessarily imply (nor have
epidemiological studies been able to confirm) any associated increase in cancer risk.
The RBE also depends on the total dose and the dose rate. Despite these limitations, the
757
Alpha
Particles
8
Overkill
7
6
RBE
1 keV
electrons
4
3
2
Diagnostic
X rays
Co-60
gamma rays
Fast
Neutrons
0.1
10
100
1000
LET (keV/m of tissue)

FIGURE 20-3 The RBE of a given radiation is an empirically derived value that, in general (with all other
factors being held constant), increases with the LET of the radiation. However, beyond approximately 100 keV/mm,
the radiation becomes less efficient due to overkill (i.e., the maximal potential damage has already been
reached), and the increase in LET beyond this point results in wasted dose.
RBE is a useful radiobiologic tool that helps to characterize the potential damage from
various types of and energies of ionizing radiation. The RBE is an essential element in
establishing the radiation weighting factors (wR) discussed in Chapter 3.
20.3 Molecular and Cellular Response to Radiation

Although all critical lesions responsible for cell killing have not been identified, it has been
established that the radiation-sensitive targets are located in the nucleus and not in the
cytoplasm of the cell. Cells contain numerous macromolecules, only some of which are
essential for cell survival. For example, there are many copies of various enzymes within
a cell; the loss of one particular copy would not significantly affect the cells function or
survival. However, if a key molecule, for which the cell has no replacement (e.g., DNA),
is damaged or destroyed, the result may be cell death. In the context of diagnostic x-ray
exposure, cell death does not mean the acute physical destruction of the cell by radiation
but rather a radiation-induced loss of mitotic capacity (i.e., reproductive death) or premature activation of apoptotic pathways (i.e., programmed cell death). There is considerable
evidence that damage to DNA is the primary cause of radiation-induced cell death.
Radiation-Induced DNA Damage and Response

Spectrum of DNA Damage
The deposition of energy (directly or indirectly) by ionizing radiation induces chemical changes in large molecules that may then undergo a variety of structural changes.
These structural changes include (1) hydrogen bond breakage, (2) molecular
758
egradation or breakage, and (3) intermolecular and intramolecular cross-linking.

d
The rupture of the hydrogen bonds that link base pairs in DNA may lead to irreversible changes in the secondary and tertiary structure of the molecule that compromise genetic transcription and translation. Molecular breakages also may involve
the sugar-phosphate polymers that comprise the backbones of the two helical DNA
strands. They may occur as single-strand breaks (SSBs), double-strand breaks (DSBs)
(in which both strands of the double helix break simultaneously at approximately the
same nucleotide pair), base loss, base changes, or cross-links between DNA strands
or between DNA and proteins. A SSB between the sugar and the phosphate can
rejoin, provided there is no opportunity for the broken portion of the strands to
separate. While the rejoining is not typically immediate, because the broken ends
require the action of a series of enzymes (endonuclease, polymerase, ligase) to rejoin,
the rejoining is fast and the repair typically occurs with high fidelity. The presence of
oxygen potentiates the damage by causing peroxidation of a base, which then undergoes radical transfer to the sugar, causing damage that prevents rejoining.
A DSB can occur if two SSBs are juxtaposed or when a single, densely ionizing
particle (e.g., an alpha particle) produces a break in both strands. DNA DSBs are very
genotoxic lesions that can result in chromosome aberrations. The genomic instability resulting from persistent or incorrectly repaired DSBs can lead to carcinogenesis
through activation of oncogenes, inactivation of tumor suppressor genes, or loss of
heterozygosity. SSBs (caused in large part by the OH radical) are more easily repaired
than DSBs and are more likely to result from the sparse ionization pattern that is
characteristic of low-LET radiation. For mammalian cells, an absorbed dose of one
Gy from x-rays will cripple the mitotic capability of approximately half of the cells
exposed. Each cell would experience approximately 40 DSBs, 1,000 SSBs, and 3,000
damaged bases. While DSBs and complex DNA damage are often associated with
high-LET radiation, in reality, all ionizing radiation is capable of producing a substantial number of complex DSBs. In the case of low-LET radiations, used in diagnostic imaging, about a quarter to a third of the absorbed dose in tissue is deposited
via low-energy secondary electrons with energies on the order of 0.1 to 5 keV. These
low-energy electrons produce high ionization densities over very short tracks that
are of the same scale as the DNA double helix. The result is an increased probability
of complex DNA damage that may contain not only SSBs and DSBs but localized
base damage as well. These complex DNA lesions are less likely to be repaired correctly than an isolated SSB, DSB, or base damage which may lead to permanent
DNA modifications or losses (Goodhead, 1988, 1994). The increased effectiveness
of alpha particles in producing biological damage is not due to an increased yield of
DNA damage but rather the ability of the higher ionization density to produce more
complex DNA lesions (Brenner and Ward, 1992). This ability to produce several
MDS in proximity in the chromatin structure is referred to as regional multiply damaged sites (RMDS). These lesions are repaired more slowly, if at all, and may serve
as a signal for gene induction for a longer time than following low-LET irradiation
(Lbrich et al., 1996). In addition, the production of RMDS increases the probability
that short double-stranded oligonucleotides will be released, making high fidelity
repair without the loss of sequence information problematic. Figure20-4 illustrates
some of the common forms of damage to DNA.
Regardless of its severity or consequences, the loss or change of a base is considered a type of mutation. Although mutations can have serious implications, changes in
the DNA are discrete and do not necessarily result in structural changes in the chromosomes. However, chromosome breaks produced by radiation do occur and can be
observed microscopically during anaphase and metaphase, when the chromosomes
A T
G
A
Base
Damage
A T
B
Base
Deletion
A T
C
Base
Substitution
A T
T A
T A
T A
T A
T C
G C
T A
T A
T A
T A
C G
C G
C G
C G
A T
T A
G C
E
Single
Strand
Break
A T
T A
G C
F
Double
Strand
Break
A T
T A
759
D
Hydrogen Bond
Disruption
G
Complex
Damage
T A
T A
C G
C G
C G
FIGURE 20-4 Several examples of DNA damage.
are condensed. Radiation-induced chromosomal lesions can occur in both somatic and
germ cells and, if not repaired before DNA synthesis, may be transmitted during mitosis
and meiosis. Chromosomal damage that occurs before DNA replication is referred to as
chromosome aberrations, whereas that occurring after DNA synthesis is called chromatid
aberrations. Unlike chromosomal aberrations, in chromatid aberrations only one of the
daughter cells will be affected if only one of the chromatids of a pair is damaged.
DNA Repair
Even in the absence of radiation, DNA damage is a relatively common event in the
life of a cell. Mammalian cells experience many thousands of DNA lesions per cell
per day as a result of a number of common cellular functions such as respiration
(e.g., oxidative damage induced by ROS during metabolism) and mitosis (e.g., errors
during base replication). Yet the mutation rate is surprisingly low due to the effectiveness of the cells response and the varied and robust DNA repair mechanisms operating
within the cell. DNA damage induces several cellular responses that enable the cell
either to repair or to cope with the damage. For example, the cell may activate the G1/S
checkpoint (which arrests cell cycle progression), to allow for repair of damaged or
incompletely replicated chromosomes. In the case of potentially catastrophic mutations,
the cell may initiate apoptosis (programmed cell death), effectively eliminating the damaged genetic material. Apoptosis is a form of cell death that is characteristically different from cell necrosis in morphology and biochemistry, leading to the elimination of
cells without releasing inflammatory substances into the surrounding area. Apoptosis
results in cell shrinkage via nuclear condensation and extensive membrane blebbing
760
ultimately resulting in the fragmentation of the cell into membrane bound apoptotic
bodies composed of cytoplasm and tightly packed organelles that are eliminated by
phagocytosis. The checkpoint and apoptosis responses utilize many of the same sensor
molecules or complexes involved in DNA damage recognition and signal transduction.
Many types of DNA repair mechanisms exist, including direct repair of a damaged
nucleotide, short and long patch base excision repair (SP BER, LP BER), nucleotide
excision repair (NER), SSB and DSB repair, and cross-link repair. The repair of DNA
damage depends on several factors, including the stage of the cell cycle and the type and
location of the lesion. For example, if there is base damage on a single strand, enzymatic
C C
C
G
C C T
C
A C
T C G G
G C
G G
C CA
G G
C C T
G G A G C C
C G G
G G A G C C
C C
T G G
C T C G G A
C
G
C
G
C C
C G G
C G G
G G A G C C
C
T G G
C T C G G A
T G G
C T C G G A
C C
A
T
C G G
C G G
C
G
G C
C C T
T C G G
G G A G C C
C
G G A G C C
C
C
A C
T G G
C T C G G A
C C T
T G G
C T C G G A
C C
G G
C CA
G G
FIGURE 20-5 A. A schematic of short patch base excision repair (SP BER), which is one of several DNA
repair pathways. (1) A cytosine base is damaged by one of several possible agents such as ROS or directly
ionizing radiation. (2,3) A DNA glycosylase detects the damaged base and removes it, creating an abasic site.
(4) A specific endonuclease is brought to the damage site where it cleaves the segment of the sugar-phosphate
backbone resulting in a SSB that can then be processed by short-patch repair. (5) DNA polymerase accompanied by the scaffolding protein resynthesizes the DNA segment using the complementary base as a template
(6) DNA ligase joins the DNA segments. (7) At this point the BER is complete and the DNA resumes its normal
helical shape. B. Scientists have recently been able to visualize the complicated and dynamic structures of DNA
ligase using a combination of x-ray crystallography and small angle x-ray scattering techniques. These experiments revealed the crystal structure of the human DNA ligase I protein bound to a short DNA oligonucleotide.
The ring-shaped structure in the center of the figure is the solvent-accessible surface of the protein. The
extended, chromosomal DNA (long coils) are an artists representation of the high-level organization of DNA
structure. The figure illustrates the ring-shaped ligase protein sliding along the DNA searching for a break in
the phosphodiester backbone of the DNA that is the substrate for the enzymes DNA end joining activity. The
enzyme, DNA ligase, repairs millions of DNA breaks generated during the normal course of a cells life, for
example, linking together the abundant DNA fragments formed during replication of the genetic material in
dividing cells. DNA ligase switches from an open, extended shape to a closed, circular shape as it joins DNA
strands together. (Courtesy of Dr. Tom Ellenberger, Department of Biochemistry and Molecular Biophysics at
Washington University School of Medicine, St. Louis. MO.)
Control (0 mGy)
200 mGy at 3min
761
200 mGy at 24h
C
FIGURE 20-5 (Continued) C. DSB induction and repair in primary human fibroblasts. Using immunofluorescence techniques, a fluorescent antibody specific for g-H2AX (a phosphorylated histone) forms discrete
nuclear foci that can be visualized at sites of DSBs. DSB repair was evaluated at 3 minutes and 24 hours
after exposure to 200 mGy; repair was almost complete at 24 hours. The length of the white scale bar
shown in the unirradiated control panel equals 10 mm. (From Rothkamm K, Lbrich M. Evidence for a lack
of DNA double-strand break repair in human cells exposed to very low x-ray doses. Proc Natl Acad Sci U S A
2003;100:50575062.)
excision occurs, and the intact complementary strand of the DNA molecule provides
the template on which to reconstruct the correct base sequence (Fig. 20-5A,B).
There are specific endonucleases and exonucleases that, along with other proteins
or complexes of proteins, are capable of repairing damage to the DNA. Most DNA
base damage and SSBs are repaired by the base excision repair pathway. NER is the
major pathway for the repair of bulky, helix-distorting lesions such as thymine dimers produced by exposure to ultraviolet radiation. While the repair of simple DNA
lesions caused by metabolism and the actions of ROS are considered the primary
substrates for the BER pathway, these lesions can also be recognized and repaired via
the NER pathway. LP BER and NER may play a more significant role in the repair of
clustered damage where significant distortions in the DNA structure may occur. DNA
repair occurs rapidly, and approximately 90% of SSB and base damage is repaired
within an hour after the initial damage. Even with DSBs, DNA rejoining is virtually
complete with 24 hours (Fig. 20-5C).
While most DSBs are repaired correctly, a few undergo binary misrepair. DSBs
can be repaired with high fidelity via homologous recombination repair (HRR)
involving exchanges with homologous DNA strands (from sister chromatids after
replication or from homologous chromosomes). More often DSBs are repaired by the
error prone nonhomologous end-joining (NHEJ) that involves end-to-end joining of
broken strands (Fig. 20-6).
HRR can preserve the genetic integrity of the chromosome while NHEJ repair
results in loss of DNA fidelity. There is a strong force of cohesion between broken
ends of chromatin material. Interchromosomal and intrachromosomal recombination may occur in a variety of ways, yielding many types of aberrations such as rings
and dicentrics. Figure 20-7 illustrates some of the more common chromosomal
aberrations and the consequences of replication and anaphasic separation. In about
half the binary misrepair events, the two DSB lead to a translocation (Fig.20-7C).
Translocations involve large scale rearrangements and can cause potentially precarcinogenic alterations in cellular phenotype, but most do not impair cellular
survival. Binary misrepair can also result in a dicentric chromosome aberration
(Fig.20-7D), which generally destroys the clonogenic viability of the cell.
762
NHEJ does not require a homologous

template for DSB repair and usually results in
the correction of the break in an error-prone
manner.
Ku 70/80
DSB Recognition - End Binding
Essential to the NHEJ pathway is the activity

of the Ku70/Ku80 heterodimeric protein
which initiates NHEJ by binding to the free
ends of the DNA.
DNA-PK
Recruitment of DNA-PK
The Ku heterodimer recruits other NHEJ

factors such as DNA-dependent protein
kinase (DNA-PK), XRCC4, and DNA Ligase IV
to the site of injury
DNA-PK becomes activated upon DNA
binding, and phosphorylates a number of
substrates including p53, Ku, and the DNA
Ligase IV cofactor XRCC4/XLF.
Phosphorylation of these factors is believed
to further facilitate the repair process.
End Processing
Ligase IV/XRCC4/XLF
Strand Synthesis & Ligation

Error Prone
Repair
Most DSBs generated by genotoxic agents are

damaged and unable to be directly ligated, they often
have to undergo limited processing by nucleases
and/or polymerases before NHEJ can proceed.
NHEJ is the simplest, fastest and most predominant

mechanism for DSB repair in humans. However, it is not
as accurate as other forms of DSB DNA repair and often
results in small deletions and insertions at the repair site.
FIGURE 20-6 Model of the key steps required for NHEJ of two ends of double strand DNA break. This repair
greatly improves the cells chance of successful replication, and any loss of DNA fidelity is limited and thus has
a low probability of compromising regulatory control or gene expression. (Adapted from Helleday T, Lo J, van
Gent DC, et al. DNA double-strand break repair: From mechanistic understanding to cancer treatment. DNA
Repair. (March 14, 2007) See animation at http://web.mit.edu/engelward-lab/animations/NHEJ.html)
Chromosomal Aberrations
The extent of the total genetic damage transmitted with chromosomal aberrations
depends on a variety of factors, such as the cell type, the number and kind of genes
deleted, and whether the lesion occurred in a somatic or in a gametic cell.
Chromosomal aberrations are known to occur spontaneously. In certain circumstances the scoring of chromosomal aberrations in human lymphocytes can be used
as a biologic dosimeter to estimate the dose of radiation received after an accidental
exposure. Lymphocytes are cultured from a sample of the patients blood and then
stimulated to divide, allowing a karyotype to be performed. The cells are arrested at
metaphase, and the frequency of rings and dicentrics are scored. Whole-body doses
from penetrating radiation in excess of 250 mGy for acute exposure and 400 mGy
for chronic exposure can be detected with confidence limits that do not include
zero. Although many chromosomal aberrations are unstable and gradually lost from
circulation, this assay is generally considered the most sensitive method for estimating recent exposure (i.e., less than 6 months). More persistent, stable reciprocal translocations can be measured using fluorescence in situ hybridization (FISH).

Breakage
A. One break in one
chromosome
B. Two breaks in one

chromosome
Rings
Recombination
Replication
763
Anaphasic Separation
None
or
C. One break in two

chromosomes
Translocation
D. One break in two

chromosomes
Dicentrics
FIGURE 20-7 Several examples of chromosomal aberrations and the effect of recombinations, replication,
and anaphasic separation. A. Single break in one chromosome, which results in centric and acentric fragments. The acentric fragments are unable to migrate and are transmitted to only one of the daughter cells.
These fragments are eventually lost in subsequent divisions. B. Ring formation may result from two breaks in
the same chromosome in which the two broken ends of the centric fragment recombine. The ring-shaped
chromosome undergoes normal replication, and the two (ring-shaped) sister chromatids separate normally at
anaphaseunless the centric fragment twists before recombination, in which case the sister chromatids will
be interlocked and unable to separate. C. Translocation may occur when two chromosomes break and the
acentric fragment of one chromosome combines with the centric fragment of the other and vice versa, or
D. the two centric fragments recombine with each other at their broken ends, resulting in the production of
a dicentric. E. Metaphase spread, containing a simple dicentric interchange between chromosomes 2 and 8
visualized with multiplex fluorescence in situ hybridization (mFISH). This technique utilizes fluorescently labeled
DNA probes with markers specific to regions of particular chromosomes which allows for the identification of
each homologous chromosome pair by its own color. The color is computer generated based on differences in
fluorescence wavelength among probes. (From Cornforth MN, et al. J Cell Biol 2002;159:237244).
In this method, chromosomes are labeled with chromosome-specific fluorescent

DNA probes, allowing translocations to be identified using fluorescent microscopy
(Fig. 20-7E). Reciprocal translocations are believed to persist for a considerable
period after the exposure, and this approach has been used as one of the methods
to estimate the doses to survivors of the atomic bombs detonated in Hiroshima and
Nagasaki decades ago.
Response to Radiation at the Cellular Level

There are a number of potential responses at the cellular level following radiation exposure. Depending on a variety of inherent and conditional biologic variables related
to the cell and its environment (e.g., cell type, oxygen tension, cell cycle at time
of exposure) as well as a number of physical factors related to the radiation exposure
(e.g., dose, dose rate, LET) a number of responses are possible such as delayed cell division,
apoptosis, reproductive failure, genomic instability (delay expression of radiation damage), DNA mutations, phenotypic (including potentially oncogenic) transformations,
bystander effects (damage to neighboring unirradiated cells), and adaptive responses
764
(irradiated cells become more radioresistant). Many of these effects are discussed in
more detail below. While a wide variety of the biologic responses to radiation have been
identified, the study of radiation-induced reproductive failure (also referred to as clonogenic cell death or loss of reproductive integrity) is particularly useful in assessing the
relative biologic impact of various types of radiation and exposure conditions. The use
of reproductive integrity as a biologic effects marker is somewhat limited, however, in
that it is applicable only to proliferating cell systems (e.g., stem cells). For differentiated
cells that no longer have the capacity for cell division (e.g., muscle and nerve cells), cell
death is often defined as loss of specific metabolic functions or functional capacity.
Cell Survival Curves

Cells grown in tissue culture that are lethally irradiated may fail to show evidence
of morphologic changes for long periods; however, reproductive failure eventually
occurs. The most direct method of evaluating the ability of a single cell to proliferate
is to wait until enough cell divisions have occurred to form a visible colony. Counting
the number of colonies that arise from a known number of individual cells irradiated
in vitro and cultured provides a way to easily determine the relative radiosensitivity
of particular cell lines, the effectiveness of different types of radiation, or the effect of
various environmental conditions. The loss of the ability to form colonies as a function of radiation exposure can be described by cell survival curves.
Several mathematical methods have been developed to model biologic response
to radiation. The shape of a cell survival curve reflects the relative radiosensitivity of
the cell line and the random nature of energy deposition by radiation. Survival curves
are usually presented with the radiation dose plotted on a linear scale on the x-axis
and the surviving fraction (SF) of cells plotted on a logarithmic scale on the y-axis.
In the multitarget model, the response to radiation is defined by three parameters: the
extrapolation number (n), the quasithreshold dose (Dq), and the D0 dose (Fig. 20-8).
The D0 describes the radiosensitivity of the cell population under study. The D0
dose is the reciprocal of the slope of the linear portion of the survival curve, and it
is the dose of radiation that produces, along the linear portion of the curve, a 37%
10
Fraction of Cells Surviving
FIGURE 20-8 Typical cell

survival curve illustrating the
portions of the curve used to
derive the extrapolation number (n), the quasithreshold
dose (Dq), and the D0 dose.
1.0
Dq
0.1
Do
0.01
0.001
6
Dose (Gy)
10
765
reduction in the number of viable cells. Radioresistant cells have a higher D0 than
radiosensitive cells. A lower D0 implies less survival per dose. The D0 for mammalian
cells ranges from approximately 1 to 2 Gy for low LET radiation.
In the case of low-LET radiation, the survival curve is characterized by an initial
shoulder before the linear portion of the semilogarithmic plot. The extrapolation
number, which gives a measure of the shoulder is found by extrapolating the linear portion of the curve back to its intersection with the y-axis. The extrapolation
number for mammalian cells ranges between 2 and 10; the larger the n, the larger
the shoulder. Dq also defines the width of the shoulder region of the cell survival
curve and is a measure of sublethal damage. Sublethal damage is a concept based
on experiments that show that when radiation dose is split into two or more fractions, with sufficient time between fractions, the cell survival increases after low-LET
radiation. The presence of the shoulder in a cell survival curve is taken to indicate
that more than one ionizing event hit, on average, is required to kill a cell and the
reappearance of the shoulder when a large dose is delivered in fractions indicates that
the cells are capable of repairing sublethal damage between fractions.
The linear quadratic (LQ) model is now the most often used to describe cell
survival studies where the SF is generally expressed as
SF (D ) e
-aDbD 2
where D is the dose in Gy, a is the coefficient of cell killing that is proportional to dose
(i.e., the initial linear component on a log-linear plot) and b is the coefficient of cell killing
that is proportional to the square of the dose (i.e., the quadratic component of the survival
curve). The two constants (a and b) can be determined for specific tissues and cancers to
predict dose response. As described previously, cell killing (i.e., loss of clonogenic viability) occurs via misrepaired or unrepaired chromosome damage such as dicentric aberrations that are formed when pairs of nearby DSBs wrongly rejoin to one another. The
double helix can undergo a DSB as the result of two different mechanisms: (1) both DNA
strands are broken by the same radiation track (or event) and (2) each strand is broken
independently, but the breaks are close enough in time and space to lead to a DSB. The
linear (alpha) component of the survival curve represents the damage done by individual
radiation particle tracks and is thus independent of dose rate. While the damage is partially repairable over time, a still represents the probability of cell death due to individual,
noninteracting, particle tracks. This linear (single-hit kinetics) dose-response relationship
dominates with high-LET radiation. The quadratic (beta) component of the survival curve
represents the probability of cell death due to interactions between two or more individual
particle tracks (i.e., dominates with low-LET radiation and follows multiple-hit kinetics)
causing the curve to bend at higher doses and is sensitive to dose rate. The LQ (or alphabeta model, as many call it) is more commonly used than the previously described n-D0
model, for several reasons: the LQ model is mechanistically based, it is more useful in
radiotherapy for explaining fractionation effect differences between late responding normal tissues and early responding tissues or tumors and the LQ model seems to fit most
experimental data on human cell lines. The dose at which cell killing is equal from the
linear (aD) and quadratic (bD2) contribution is referred to as the a/b ratio. The a/b ratio
is a measure of the curvature of the cell survival curve and, thus, a measure of the sensitivity of different cell types to fractionation of radiation dose, Figure 20-9. For example, late
responding normal tissues such as spinal cord or lung that have smaller a/b ratios of 3 or
4 are preferentially spared by fractionation compared to tumors and early responding
normal tissues (gut, skin, bone marrow) where the a/b ratio is larger (8 to 12), indicating
less ability to repair (i.e., more alpha component and less effect of fractionating the dose).
766
FIGURE 20-9 The LQ model. The experimental data

are fitted to a LQ function. There are two components
to cell killing: One is proportional to dose (aD); the
other is proportional to the square of the dose bD2.
The dose at which the linear and quadratic components are equal is the ratio a/b. The LQ curve bends
continuously but is a good fit to the experimental
data for the first few decades of survival. (From Hall
EJ, Giaccia AJ. Radiobiology for the radiologist. 7th
ed. Philadelphia, PA: Lippincott Williams and Wilkins,
2012, with permission.)
Factors Affecting Cellular Radiosensitivity

Cellular radiosensitivity can be influenced by a variety of factors that can enhance
or diminish the response to radiation or alter the temporal relationship between
the exposure and a given response. These factors can be classified as either conditional or inherent. Conditional radiosensitivities are those physical or chemical factors that exist before and/or at the time of irradiation. Some of the more important
conditional factors affecting dose-response relationships are discussed in the following paragraphs, including dose rate, LET, and the presence of oxygen. Inherent
radiosensitivity includes those biologic factors that are characteristics of the cells
themselves, such as the mitotic rate, the degree of differentiation, and the stage of
the cell cycle.
Conditional Factors
The rate at which a dose of low-LET radiation is delivered has been shown to affect
the degree of biologic damage for a number of biologic end points including chromosomal aberrations, reproductive delay, and cell death. In general, high dose rates
are more effective at producing biologic damage than low dose rates. The primary
explanation for this effect is the diminished potential for repair of radiation damage.
Cells have a greater opportunity to repair sublethal damage at low dose rates than at
higher dose rates, reducing the amount of damage and increasing the survival fraction. Figure 20-10 shows an example of the dose rate effect on cell survival.
Note that the broader shoulder associated with low-dose-rate exposure indicates
its diminished effectiveness compared with the same dose delivered at a higher dose
rate. This dose-rate effect is diminished or not seen with high-LET radiation primarily because the dense ionization tracks typically produce more complex, clustered DNA damage that cannot be repaired correctly. However, for a given dose rate,
high-LET radiation is considerably more effective in producing cell damage than
low-LET radiation (Fig. 20-11).
For a given radiation dose, a reduction in radiation damage is also observed
when the dose is fractionated over a period of time. This technique is fundamental to
the practice of radiation therapy. The intervals between doses (hours to a few days)
allow the repair mechanisms in healthy tissue to gain an advantage over the tumor by
repairing some of the sublethal damage. Figure 20-12 shows an idealized experiment
767
10
Low dose rate

0.1 Gy/min
1.0
0.1
High dose rate

1 Gy/min
0.01
0.001
8
10
Dose (Gy)
12
14
16
18
FIGURE 20-10 Cell survival curves illustrating the effect of dose rate for low LET radiation. Lethality is
reduced because repair of sublethal damage is enhanced when a given dose of radiation is delivered at a low
versus a high dose rate.
in which a dose of 10 Gy is delivered either all at once or in five fractions of 2 Gy with

sufficient time between fractions for repair of sublethal damage.
For low-LET radiation, the decreasing slope of the survival curve with decreasing
dose rate (see Fig. 20-10) and the reoccurrence of the shoulder with fractionation
(see Fig. 20-12) are clear evidence of repair.
The presence of oxygen increases the damage caused by low-LET radiation by
inhibiting the recombination of free radicals to form harmless chemical species and
10
1.0
Low LET
(x-rays)
High LET
(neutrons)
0.1
0.01
High LET
(alphas)
0.001
6
Dose (Gy)
10
FIGURE 20-11 Cell survival curves illustrating the greater damage produced by radiation with high-LET. At 10%
survival, high-LET radiation is three times as effective as the same dose of low-LET radiation in this example.
768
106
D1
Number of Surviving Cells
D2
105
A
D3
104
D4
D5
103
D
E
102
Dose
FIGURE 20-12 Idealized fractionation experiment depicting the survival of a population of 106 cells as a
function of dose. Curve A represents one fraction of 10 Gy. Curve F represents the same total dose as in curve
A delivered in equal fractionated doses (D1 through D5) of 2 Gy each, with intervals between fractions sufficient to allow for repair of sublethal damage. (Modified from Hall EJ. Radiobiology for the radiologist. 5th ed.
Philadelphia, PA: Lippincott Williams & Wilkins, 2000.)
by inhibiting the chemical repair of damage caused by free radicals. This effect is
demonstrated in Figure 20-13, which shows a cell line irradiated under aerated and
hypoxic conditions.
The relative effectiveness of radiation to produce damage at various oxygen tensions is described by the oxygen enhancement ratio (OER). The OER is defined
Low LET
OER = 3.0
(e.g. x-rays)
1.0
Hypoxic
O2
0.1
0.01
0.001
4
6
Dose (Gy)
High LET
OER = 1.7
(e.g. neutrons)
10
10
10
1.0
Hypoxic
O2
0.1
0.01
0.001
4
6
Dose (Gy)
10
FIGURE 20-13 Cell survival curves demonstrating the effect of oxygen during high (blue) and low (green)
oxygen tension on the OER for high and low-LET irradiation.
769
as the dose of radiation that produces a given biologic response in the absence of
oxygen divided by the dose of radiation that produces the same biologic response in
the presence of oxygen. Increasing the oxygen concentration at the time of irradiation has been shown to enhance the killing of otherwise hypoxic (i.e., radioresistant)
cells that can be found in some tumors. The OER for mammalian cells in cultures is
typically between 2 and 3 for low-LET radiation. High-LET damage is not primarily
mediated through free radical production, and therefore the OER for high-LET radiation can be as low as 1.0.
Inherent Factors
In 1906, two French scientists, J. Bergonie and L. Tribondeau, performed a series
of experiments that evaluated the relative radiosensitivity of rodent germ cells at
different stages of spermatogenesis. From these experiments, some of the fundamental characteristics of cells that affect their relative radiosensitivities were established.
The Law of Bergonie and Tribondeau states that radiosensitivity is greatest for those
cells that (1) have a high mitotic rate, (2) have a long mitotic future, and (3) are
undifferentiated. With only a few exceptions (e.g., lymphocytes), this law provides a
reasonable characterization of the relative radiosensitivity of cells in vitro and in vivo.
For example, the pluripotential stem cells in the bone marrow have a high mitotic
rate, a long mitotic future, are poorly differentiated, and are extremely radiosensitive compared with other cells in the body. On the other end of the spectrum, the
fixed postmitotic cells that comprise the central nervous system (CNS) are relatively
radioresistant (Fig.20-14).
This classification scheme was refined in 1968 by Rubin and Casarett, who
defined five cell types according to characteristics that affect their radiosensitivity
(Table 20-1).
Most Sensitive
Spermatogonia
Lymphocytes, oocytes
Hematopoietic stem cells
Small intestine crypt cells
Hair follicles
Colon, stomach
Skin, kidney
CNS (neurons)
Muscle
Bone
Least Sensitive
FIGURE 20-14 Relative radiosensitivity of tissues.
770
TABLE 20-1 CLASSIFICATION OF CELLULAR RADIOSENSITIVITY

CELL TYPE
CHARACTERISTICS
EXAMPLES
RADIOSENSITIVITY
VIM
Rapidly dividing;
undifferentiated; do not
differentiate between
divisions
Type A spermatogonia
Erythroblasts
Crypt cells of intestines
Basal cells of epidermis
Most radiosensitive
DIM
Actively dividing; more

differentiated than VIMs;
differentiate between
divisions
Intermediate spermatogonia
Myelocytes
Relatively radiosensitive
MCT
Irregularly dividing; more

differentiated than VIMs
or DIMs
Endothelial cells
Fibroblasts
Intermediate in
radiosensitivity
RPM
Do not normally divide but Parenchymal cells of the

retain capability of division; liver and adrenal glands
differentiated
Lymphocytesa
Bone
Muscle cells
Relatively radioresistant
FPM
Do not divide;
differentiated
Most radioresistant
Some nerve cells

Erythrocytes
Spermatozoa
Lymphocytes, although classified as relatively radioresistant by their characteristics, are in fact very
radiosensitive.
VIM, vegetative intermitotic cells; DIM, differentiating intermitotic cells; MCT, multipotential connective tissue
cells; RPM, reverting postmitotic cells; FPM: fixed postmitotic cells.
Source: Rubin P, Casarett GW. Clinical radiation pathology as applied to curative radiotherapy. Clin Pathol
Radiat 1968;22:767768.
The stage of the cells in the reproductive cycle at the time of irradiation greatly
affects their radiosensitivity. Figure 20-15 shows the phases of the cell reproductive
cycle and several checkpoints that can arrest the cell cycle or interrupt the progression to allow for the integrity of key cellular process to be evaluated and if necessary,
repaired. Experiments indicate that, in general, cells are most sensitive to radiation
during mitosis (M phase) and the gap between S phase and mitosis (G2), less sensitive during the preparatory period for DNA synthesis (G1), and least sensitive during
late DNA synthesis (S phase).
Adaptive Response, Bystander Effect, and Genomic Instability

A number of other responses to radiation have been observed in vitro that raise
interesting questions about the applicability of the LQ dose-response model for lowdose low-LET radiation used in medical imaging. An adaptive response to radiation
has been demonstrated in which an initial exposure or priming dose reduced the
effectiveness of a subsequent exposure. For example, it has been demonstrated in
vitro with human lymphocytes, which compared to controls did not receive a small
initial exposure, a priming dose of 10 mGy significantly reduced the frequency of
chromosomal aberrations in the same cells exposed to several Gy a few hours later
(Shadley, 1987). However, the magnitude of this adaptive response varies considerably with dose and dose rate as well as among lymphocytes from different individuals and with other variables. Many other end points for adaptive response have
been studied such as cell lethality, mutations, and defects in embryonic development
for which the evidence for an adaptive response was highly variable. While many
771
se
pha
metaphase
anaphase
telo
pha
pro
M
Checkpoint
G2/M
Checkpoint
G2
Post DNA
Synthesis
Pre Mitosis
se
M
Mitosis
Quiescence
G1
Pre DNA
Synthesis
S
DNA
Synthesis
Inhibition
G0
Stimulation
Interphase
G1/S
Checkpoint
FIGURE 20-15 Phases of the cells reproductive cycle. The time between cell divisions is called interphase.
Interphase includes the period after mitosis but before DNA synthesis (G1), which is the most variable in length
of the phases; followed by S phase, during which DNA synthesis occurs; followed by G2, all leading up to
mitosis (M phase), the events of which are differentiated into prophase, metaphase, anaphase, and telophase.
Much of the control of the progression through the phases of a cell cycle is exerted by specific cell cycle control
genes at checkpoints. Checkpoints are critical control points in the cell cycle that have built-in stop signals that
halt the cell cycle until overridden by external chemical signals to proceed. There are three major checkpoints
in the cell cycle, G1 checkpoint between G1 and S phase (G1/S), the G2 checkpoint between G2 and mitosis
(G2/M) and the M (metaphase) checkpoint. The G1/S checkpoint is where the cell monitors its size, available
nutrients, and the integrity of DNA (e.g., prevents copying of damaged bases, which would fix mutations in
the genome) to assure all are adequate for cell division. In the absence of a proceed signal, the cell will enter a
quiescent state G0 (the state of most cells in the body) until it receives a stimulation signal to continue. Following DNA synthesis, the G2/M checkpoint occurs at the end of the G2 phase, during which DNA damage induced
during replication, such as mismatched bases and double-stranded breaks, is repaired. The cell ensures that
DNA synthesis (S-Phase) has been successfully completed before triggering the start of the mitosis, and at the
metaphase (spindle) checkpoint the cell monitors spindle formation and ensures all chromosomes are attached
to the mitotic spindle by kinetochores prior to advancing to anaphase. The tumor-suppressor gene p53 (so
named because it encodes a phosphorylated protein with a molecular weight of 53 kDa), operates at the G1/S
and G2/M checkpoints. The p53 protein (discussed again in relation to radiation-induced carcinogenesis later
in the chapter) induces cell-cycle arrest through the inhibition of cyclin dependent protein kinases (required
for cell cycle advancement) and thus allows for repair of DNA damage. This protein also activates DNA repair
mechanisms, or (in the case of severe DNA damage) induces cell death via apoptosis.
theories have been advanced to explain this phenomenon, there is still insufficient
evidence to use these results to modify the dose-response relationship for human
exposure to radiation.
The bystander effect is another fascinating phenomenon in which irradiated cells
or tissues can produce deleterious effects on nonirradiated cells or tissues. This
so-called abscopal (out-of-field) effect of radiation has been demonstrated in a variety
of experiments. One of the earliest examples was the ability of plasma from patients
who had received radiation therapy to induce chromosomal aberrations in lymphocytes
from nonirradiated patients (Hollowell, 1968). Among the most compelling evidence
for the bystander effect are in vitro experiments in which an a-particle microbeam, with
772
the ability to irradiate a single cell, can produce a host of deleterious effects in neighboring unirradiated cells. Examples of induced changes in the unirradiated cells include
evidence of DNA damage such as micronuclei formation, sister-chromatid exchanges,
cell killing, as well as up and down regulation of a number of important genes (e.g.,
p53 and rad51), genomic instability (discussed below) and even malignant transformation. Like adaptive response, the sequence of events following exposure is varied
and complex and while many molecular mechanisms have been proposed to explain
the bystander effects, relationship of this phenomenon to low-dose, low-LET radiation
effects characteristic of medical radiation exposure in humans is still an open question.
While the vast majority of unrepaired and misrepaired radiation-induced lesions
are expressed as chromosomal damage at the first division, a certain fraction of cells
can express chromosomal damage such as chromosomal rearrangements, chromatid
breaks and gaps, and micronuclei over many cell cycles after they are irradiated.
The biological significance and molecular mechanism surrounding this persistent
genomic instability has been an area of active research for many years. Genomic instability has been demonstrated in vitro as delayed lethality in which cell cloning efficiency is reduced several generations after irradiation. Another interesting aspect is
the differences in the types of mutations associated with radiation-induced genomic
instability. Experiments have shown that the unirradiated progeny of the irradiated cells primarily demonstrate a de novo increase in lethal point mutations several
generations after the initial irradiation. These mutations are more typical of spontaneous mutations than deletions and other mutations induced directly by ionizing
radiation. There is evidence that suggests that errors induced during DNA repair
may contribute to genomic instability. For example, experiments with cells deficient
in the repair enzymes needed for NHEJ repair of radiation-induced DSBs demonstrate greater genomic instability than normal cells of the same type (Little, 2003).
However, there are data to suggest that many other factors such as ROS, alterations
in signal transduction pathways, centrosome defects and other factors also play a role
in radiation-induced genomic instability. Despite the many experimental models that
have revealed different aspects of this phenomenon, the search for the relevance of
radiation-induced genomic instability to radiation-induced cancer continues.
While radiation-induced responses such as genomic instability, adaptation,
bystander effects, (as well as others which have not been discussed such as low-dose
hypersensitivity) are fascinating in their own right, the results obtained are often
restricted to specific experimental conditions and clear mechanistic understanding about these phenomena is still lacking. It has been suggested that these effects
may alter the responses of cells and tissues to low doses of radiation, especially for
carcinogenesis induction; however, at this time, they cannot be reliably used as a
modifying factor to predict the biological consequences of radiation exposure in
humans. On the other hand, the nonlinear nature of these and other multicellular and
tissue-level responses raise serious questions regarding the current paradigm of linear extrapolation of risk based on the individual cell and the target. Addressing these
complex responses to radiation interactions requires a multidimensional, systemslevel approach and is an active area of current radiobiological research.
20.4 Organ System Response to Radiation

The response of an organ system to radiation depends not only on the dose, dose
rate, and LET of the radiation but also on the relative radiosensitivities of the cells
that comprise both the functional parenchyma and the supportive stroma. In this
773
case, the response is measured in terms of morphologic and functional changes

of the organ system as a whole rather than simply changes in cell survival and
kinetics.
The response of an organ system after irradiation occurs over a period of time.
The higher the dose, the shorter the interval before the physiologic manifestations of
the damage become apparent (latent period) and the shorter the period of expression
during which the full extent of the radiation-induced damage is evidenced. There
are practical threshold doses below which no significant changes are apparent. In
most cases, the pathology induced by radiation is undistinguishable from naturally
occurring pathology.
Regeneration and Repair

Healing of tissue damage produced by radiation occurs by means of cellular regeneration (repopulation) and repair (Fig. 20-16). Regeneration refers to replacement of the
damaged cells in the organ by cells of the same type, thus replacing the lost functional capacity. Repair refers to the replacement of the damaged cells by fibrotic scar
tissue, in which case the functionality of the organ system is compromised. The types
of response and the degree to which they occur are functions of the dose, the volume
of tissue irradiated, and the relative radiosensitivity and regenerative capacity of the
cells that comprise the organ system. In so far as repopulation at the cellular level
occurs within days after irradiation (Trott, 1991) fractionation of the dose (e.g., multiple fluoroscopically guided interventional procedures separated by days or weeks)
allows for cellular repair and typically results in less extensive tissue damage than if
the same total dose were to be delivered all at once. In radiation therapy, fractionation increases the effectiveness of the radiation by allowing for reoxygenation of the
tumor cells and reassortment (redistribution) of the irradiated cells into more radiosensitive phases of the cell cycle. If the exposures are excessive, the ability of the cells
to affect any type of healing may be lost, resulting in tissue fibrosis and necrosis.
Radiation dose
ORGAN SYSTEM
Dose too
high
Tissue Necrosis
Healing
Repair
Regeneration
Replacement of damaged cells

by different cell type (i.e. fibrosis)
Replacement of damaged cells

by same cell type
Organ is not returned to its

preirradiated state
Organ returned to preirradiated

state
Radioresistant tissues (FPM cell

types) can only effect repair e.g.,
muscle and CNS
Radiosensitive tissues (VIM,

DIM, MCT) e.g., skin, G.I. tract,
bone marrow
FIGURE 20-16 Schematic diagram of organ system response to radiation.
774
Characterization of Radiosensitivity
The characterization of an organ system as radioresistant or radiosensitive depends
in large part on the radiosensitivity of cells that comprise the functional parenchyma. The cells of the supportive stromal tissue consist mainly of cells of intermediate radiosensitivity. Therefore, when the parenchyma contains radiosensitive
cell types (e.g., stem cells), the initial hypoplasia and concomitant decrease in functional integrity will be the result of damage to these radiosensitive cell populations.
Functional changes are typically apparent within days or weeks after the exposure.
However, if the parenchyma is populated by radioresistant cell types (e.g., nerve
or muscle cells), damage to the functional layer occurs indirectly by compromise
of the cells in the vascular stroma. In this case, the hypoplasia of the parenchymal
cells is typically delayed several months and is secondary to changes in the supportive vasculature. Late radiation effects on the vasculature include fibrosis, proliferation of myointimal cells, and hyaline sclerosis of arterioles, the effect of which
is a gradual narrowing of the vessels and reduction in the blood supply to the point
that the flow of oxygen and nutrients is insufficient to sustain the cells comprising the functional parenchyma (Fig. 20-17). The relative radiosensitivity of various
tissue and organs and the primary mechanism for radiation-induced parenchymal
hypoplasia is shown in Table 20-2.
Specific Organ System Responses

This section focuses on radiation-induced changes to the skin, reproductive organs,
and eyes. Effects on the hematopoietic, gastrointestinal, and cardiovascular systems,
and the CNS are addressed in the context of the acute radiation syndrome (ARS).
Additional information on radiation effects on these and other tissues and organ
systems can be found in several textbooks and review publications listed at the end
of the chapter under sections for suggested reading and references (e.g., Hall, 2012;
ICRP, 2011; Mettler, 2008).
Skin
While radiation-induced skin damage is a relatively rare event, it is still the most
commonly encountered tissue reaction (deterministic effect) following high-dose
image guided interventional procedures. Acute radiation-induced skin changes were
recognized soon after the discovery of x-rays and were reported in the literature as
early as 1896 (Codman, 1896; Daniel, 1896). The first evidence of biologic effects
FIGURE 20-17 Histopathology showing

radiation-induced arteriole fibrosis (left).
(From Zaharia M, Goans RE, Berger ME, et
al. Industrial radiography accident at the
Yanango hydroelectric power plant. In:
Ricks RC, et al., eds. The medical basis for
radiation-accident preparedness, the clinical care of victims. New York: The Parthenon Publishing Group, 2001:267281.)
775
TABLE 20-2RELATIVE ORGAN AND TISSUE RADIOSENSITIVITY
AND PRIMARY MECHANISM FOR RADIATION-INDUCED

PARENCHYMAL HYPOPLASIA
RELATIVE
RADIOSENSITIVITY
CHIEF MECHANISM OF
PARENCHYMAL HYPOPLASIA
Lymphoid organs; bone

marrow; testes and ovaries;
small intestines
High
Destruction of parenchymal cells,

especially the vegetative and
differentiating cells
Skin and other organs with

epithelial cell lining (cornea,
lens, oral cavity, esophagus,
GI organs, bladder, vagina,
uterine cervix, uterus, rectum)
Fairly high
Destruction of radiosensitive vegetative

and differentiating parenchymal cells of
the epithelial lining
Growing cartilage; the

vasculature; growing bones
Medium
Destruction of proliferating
chondroblasts or osteoblasts; damage
to the endothelium; destruction of
connective tissue cells and chondroblasts
or osteoblasts
Mature cartilage or bone;

lungs; kidneys; liver;
pancreas; adrenal gland;
pituitary gland; thyroid;
salivary glands
Fairly low
Hypoplasia secondary to damage to the

fine vasculature and connective tissue
elements
Muscle; brain; spinal cord
Low
Hypoplasia secondary to damage to the

fine vasculature and connective tissue
elements, with little contribution by the
direct effect on parenchymal tissues
ORGANS
Note: Cells of the testes are more sensitive than ovaries. Skin radiosensitivity is particularly high around the
hair follicles.
Source: Adapted from Rubin P, Casarett GW. Clinical radiation pathology. Philadelphia, PA: W.B. Sanders, 1968.
of ionizing radiation appeared on exposed skin in the form of erythema and acute
radiation dermatitis. In fact, before the introduction of the roentgen as the unit of
radiation exposure, radiation therapists evaluated the intensity of x-rays by using a
quantity called the skin erythema dose, which was defined as the dose of x-rays
necessary to cause a certain degree of erythema within a specified time. This quantity
was unsatisfactory for a variety of reasons, not least of which was that the response to
skin from radiation exposure is quite variable. Malignant skin lesions from chronic
radiation exposure were reported as early as 1902 (Frieben, 1902).
The reaction of skin to ionizing radiation (often referred to as the cutaneous radiation syndrome) has been studied extensively, and the degree of damage has been found
to depend not only on the radiation quantity, quality, and dose rate but also on the
location and extent of the exposure. Radiation-induced skin injuries can be severe;
debilitating, and in some cases, the dose has been high enough to cause chronic
ulceration and necrosis requiring surgical intervention and a prolonged course of
intense care lasting years.
While radiation oncologists are well versed in the potential for skin injury
from radiotherapy, most physicians (including radiologists) are unfamiliar with
the appearance, time course, and doses necessary to produce clinically significant skin damage. There are often no immediate clinical symptoms from a
high skin dose and when initial symptoms do develop (e.g., erythema, xerosis,
776
ruritus) they are often delayed by weeks or even months. Primary care physip
cians evaluating their patients often fail to consider the patients past radiologic
procedure as a potential cause of their symptoms. Fortunately, serious skin injuries are rare and with the exception of prolonged, fluoroscopically guided interventional procedures and rare cases where excessive doses have been received
from CT it is highly unlikely that the radiation dose from carefully performed
(i.e., optimized) diagnostic examinations will be high enough to produce any of
the effects discussed below.
Skin damage is a consequence of acute radiation-induced oxidative stress resulting in a cascade of inflammatory responses, reduction and impairment of functional
stem cells, endothelial cell changes, and epidermal cell death via apoptosis and
necrosis. The most sensitive structures in the skin include the germinal epithelium,
sebaceous glands, and hair follicles. The cells that make up the germinal epithelium,
located between the dermis and the epidermis, have a high mitotic rate and continuously replace sloughed epidermal cells. Complete turnover of the epidermis normally
occurs within approximately 4 to 6 weeks. Skin reactions to radiation exposure are
deterministic and have a threshold of approximately 1 Gy below which no effects are
seen. At higher doses, radiation can interfere with normal maturation, reproduction,
and repopulation of germinative epidermal cell populations. At very high doses, the
mitotic activity in the germinal cells of the sebaceous glands, hair follicles, basal cell
layer, and intimal cells of the microvasculature can be compromised.
A generalized erythema can occur within hours following an acute dose of
2 Gy or more of low-LET radiation and will typically fade within a few hours or
days. This inflammatory response, often referred to as early transient erythema,
is largely caused by increased capillary dilatation and permeability secondary to
the release of vasoactive amines (e.g., histamine). Higher doses produce earlier
and more intense erythema. A later wave of erythema can reappear as early as 2
weeks after a high initial exposure or after repeated lower exposures (e.g., 2 Gy/
day as in radiation therapy), reaching a maximal response about the third week, at
which time the skin may be edematous, tender, and often exhibit a burning sensation. This secondary or main erythema is believed to be an inflammatory reaction
secondary to release of proteolytic enzymes from damaged epithelial basal cells
as well as reflecting loss of those epithelial cells. The oxidative stress resulting
from a burst of radiation-induced free radicals is known to upregulate numerous
pathways pertinent to vascular damage, including adhesion molecules, proinflammatory cytokines, smooth muscle cell proliferation, and apoptosis. A third or late
erythema wave may also be seen between between 8 and 52 weeks after exposure.
The dermal ischemia present at this stage produces an erythema with a bluish or
mauve tinge.
Temporary hair loss (epilation) can occur in approximately 3 weeks after exposure to 3 to 6 Gy, with regrowth beginning approximately 2 months later and complete within 6 to 12 months. After moderately large doses40 Gy over a period of
4 weeks or 20 Gy in a single doseintense erythema followed by an acute radiation dermatitis and moist desquamation occurs and is characterized by edema, dermal hypoplasia, inflammatory cell infiltration, damage to vascular structures, and
permanent hair loss. Moist desquamation, which implies a total destruction of the
epidermis, is a clear predictor of late delayed injuries, particularly telangiectasia.
Provided the vasculature and germinal epithelium of the skin have not been too
severely damaged, reepithelialization occurs within 6 to 8 weeks, returning the
skin to normal within 2 to 3 months. If these structures have been damaged but
not destroyed, healing may occur although the skin may be atrophic, hypo- or
777
hyper-pigmented, and easily damaged by minor physical trauma. Recurring lesions

and infections at the site of irradiation are common in these cases, and necrotic ulceration can develop. Chronic radiation dermatitis can also be produced by repeated
low-level exposures (10 to 20 mGy/d) where the total dose approaches 20 Gy or more.
In these cases the skin may become hypertrophic or atrophic and is at increased risk
for development of skin neoplasms (especially squamous cell carcinoma). Erythema
will not result from chronic exposures in which the total dose is less than 6 Gy.
The National Cancer Institute (NCI) has defined four grades of radiationinduced skin toxicity where grade 1 is the least severe and grade 4 is the most
severe. The range or band of doses associated with each grade and the anticipated skin damage as well as the temporal character of the response is shown in
Table 20-3. Figure 20-18 illustrates several grades of radiation-induced skin reactions from exposure to diagnostic and interventional imaging procedures.
Skin contamination with radioactive material can also produce skin reactions.
The extent of the reaction will depend on the quantity of radioactive material,
characteristics of the radionuclide including the types and energy of the radiations
emitted, the half-life and specific activity, region of the skin that was contaminated
and how long the contamination remained on the skin. Radionuclides with high
specific activity and high-energy beta particle emissions such as F-18 are more
likely to illicit skin reactions (see Figure 20-18B).
For all end points, the higher the dose and dose rate (beyond the threshold for
effects) the shorter the latency and more severe the effect. However, it is important to
recognize that the dose ranges shown in Table 20-3 are not to be interpreted as clear
demarcations between various skin reactions and their associated dose. There are a
number of factors that may cause the individual patient to be more or less sensitive
to radiation exposure. Biologic factors, such as diabetes mellitus, systemic lupus
erythematous, scleroderma, or mixed connective tissue disease, and homozygosity
for ataxia-telangiectasia (A-T), have increased sensitivity and potential for severe
skin reactions. Other physical and biological variables that can substantially modify
the severity of radiation-induced skin damage include size of the exposure area,
anatomical location, fractionation, patient health, and medications (Table 20-4).
Reproductive Organs
In general, the gonads are very radiosensitive. The testes contain cell populations
that range from the most radiosensitive germ cells (i.e., spermatogonia) to the most
radioresistant mature spermatozoa. The other cell populations with progressively
greater differentiation during the 10-week maturation period (i.e., primary and secondary spermatocytes and spermatids) are of intermediate radiosensitivity compared
to the germ cells and mature sperm. The primary effects of radiation on the male
reproductive system are reduced fertility, temporary, and permanent sterility (azoospermia) (Clifton and Bremner, 1983). Temporary and permanent sterility can occur
after acute doses of approximately 500 mGy and 6 Gy, respectively. The duration of
temporary sterility is dose dependent with recovery beginning at 1 and as long as
3.5 years after doses of 1 and 2 Gy, respectively. However, following exposure (and
provided the dose is not excessive), there will be a window of fertility before the
onset of sterility, as long as mature sperms are available. Chronic exposures of 20 to
50 mGy/wk can result in permanent sterility when the total dose exceeds 2.5 to 3 Gy.
The reduced threshold for affect following chronic versus acute exposure is unusual
(i.e., an inverse fractionation effect) and is believed to be due to stem cells progressing into radiosensitive stages (Lushbaugh and Ricks, 1972). Reduced fertility due to
decreased sperm count (oligospermia) and motility (asthenozoospermia) can occur
Transient erythema
Transient erythema
Erythema, epilation Dermal atrophy
After very high doses: edema
Moist desquamation Secondary ulceration due to
and acute ulceration, long-term
failure of moist desquamation
surgical intervention likely to be
to heal, surgical intervention
required
likely to be required
At higher doses: dermal
necrosis, surgical intervention
likely to be required
34
.15
Telangiectasiag
Dermal atrophy induration
Possible late skin
breakdown
Wound might be persistent
and progress into a deeper
lesion
Surgical intervention likely
to be required
Telangiectasiag
Dermal atrophy induration
Skin likely to be weak;
atrophic
Recovery
At higher doses: dermal
atrophy induration
None expected
.40 wk
LONG
This table is applicable to the normal range of patient radiosensitivities in the absence of mitigating or aggravating physical or clinical factors.
This table does not apply to the skin of the scalp.
c
Skin dose refers to actual skin dose (including backscatter). This quantity is not air kerma at the reference point (Ka,r).
d
Skin dosimetry based on Ka,r or PKA is unlikely to be more accurate than 50%.
e
The dose range and approximate time period are not rigid boundaries. Also, signs and symptoms can be expected to appear earlier as the skin dose increases.
f
Abrasion or infection of the irradiated area is likely to exacerbate radiation effects.
g
Refers to radiation-induced telangiectasia. Telangiectasia associated with an area of initial moist desquamation or the healing of ulceration may be present earlier.
Source: Radiation dose management for fluoroscopically-guided interventional procedures. NCRP Report No. 168. Bethesda, MD: National Council on Radiation Protection, 2010.
Erythema, epilation Prolonged erythema

Possible dry or moist Permanent epilation
desquamation
Recovery from
desquamation
Recovery
At higher doses: prolonged
erythema, permanent partial
epilation
Recovery from hair loss
23
Erythema, epilation
Epilation
1015
Transient erythema
12
510
Transient erythema
25
652 wk
28 wk
No observable effects expected at any time
MID TERM
EARLY
Not
applicable
PROMPT ,2 wk
APPROXIMATE TIME OF ONSET OF EFFECTSe,f
02
SINGLE-SITE
NCI (2006)
ACUTE SKIN-DOSE SKIN REACTION
RANGE (Gy)ce
GRADE
BUTTOCKS, OR ARMSa,b
TABLE 20-3TISSUE REACTIONS FROM A SINGLE-DELIVERY RADIATION DOSE TO THE SKIN OF THE NECK, TORSO, PELVIS,
778
779
FIGURE 20-18 Examples of radiation-induced effects on skin. A. National Cancer Institute (NCI) skin toxicity grade 1: Two fluoroscopically guided procedures were performed through overlapping skin ports in a
65-year-old man. Note enhanced reaction in the overlap zone. The first procedure was performed 6 weeks
before and the second procedure, 2 weeks before this photograph was obtained (Balter S, Hopewell JW,
Miller DL, et al. Fluoroscopically guided interventional procedures: a review of radiation effects on patients
skin and hair. Radiology 2010;254(2):326341). B. Radionuclide skin contamination. Slight erythema after
accidental contamination by F-18 fluorodeoxyglucose (https://rpop.iaea.org/rpop/rpop/content/informationfor/healthprofessionals/5_interventionalcardiology/phaseserythema.htm, Accessed July 11, 2011). C. Epilation
following unnecessarily high-dose CT perfusion scan. (Photo courtesy of New York Times: The Radiation Boom
After Stroke Scans, Patients Face Serious Health Risks by Walt Bogdanich Published: July 31, 2010.) D. NCI
skin toxicity grade 3. Chronic radiodermatitis after interventional cardiac catheterization. Grade 3 is classified
as moist desquamation in areas other than skinfolds and creases. Note the increased severity of reaction in an
area of radiation field overlap (Henry FA, et al. Fluoroscopy-induced chronic radiation dermatitis: A report of
three cases. Dermatol Online J 2009). E. Dry desquamation (poikiloderma) at one month in a patient receiving approximately 11 Gy calculated peak skin dose (Chambers C, Fetterly K, Holzer R, et al. Radiation safety
program for the cardiac catheterization laboratory. Catheter Cardiovasc Interv. 2011;77). FH. NCI skin toxicity
grade 4. A 40-year-old male who underwent multiple coronary angiography and angioplasty procedures. The
photographs show the time sequence of a major radiation injury (Shope TB. Radiation-induced skin injuries
from fluoroscopy. RadioGraphics 1996;16:11951199). (F) Six to eight weeks postexposure (prolonged erythema with mauve central area, suggestive of ischemia). The injury was described as turning red about 1
month after the procedure and peeling a week later. By 6 weeks, it had the appearance of a second-degree
burn; (G) sixteen to twenty-one weeks postexposure (depigmented skin with central area of necrosis); and (H)
eighteen to twenty-one months postexposure (deep necrosis with atrophic borders). Skin breakdown continued over the following months with progressive necrosis. The injury eventually required a skin graft. While the
magnitude of the skin dose received by this patient is not known, from the nature of the injury it is probable
that the dose exceeded 20 Gy. This sequence is available on the FDA Web site (NCRP 168, 2010).
780
ACUTE EFFECTS
(AFTER DAYS - WEEKS)
NORMAL ANATOMY
dry desquamation
ulceration
hair
pigmentation
epidermis
moist desquamation
sweat gland
erythema and edema
dermis
capillaries
blood vessels
laceration
subcutaneous fat
sebaceous gland
lymph vessels
CHRONIC EFFECTS
(AFTER MONTHS - YEARS)
secondary cancer
ulceration
depigmentation pigmentation
fibrosis
necrosis
atrophy of :
sweat and sebaceous glands
regenerating skin cells
hair follicles
capillaries
lymph vessels
ZA NI
O CH
RS O
KY LA
,M S
.D
.
telangiectasia
FIGURE 20-18 (Continued) I. A three dimensional view depicting the spectrum of radiation-induced effects
on skin as shown in the previous photos (Figures 20-18AH) and discussed in the text. (Source: Courtesy of
Nicholas Zaorsky, MD.)
6 weeks after a dose of 150 mGy. These effects are not related to diagnostic examinations, because acute gonadal doses exceeding 100 mGy are unlikely.
The ova within ovarian follicles (classified according to their size as small, intermediate, or large) are sensitive to radiation. The intermediate follicles are the most
radiosensitive, followed by the large (mature) follicles and the small follicles, which
are the most radioresistant. Therefore, after a radiation dose as low as 1.5 Gy, fertility may be temporarily preserved owing to the relative radioresistance of the mature
follicles, and this may be followed by a period of reduced fertility. Fertility will recur
provided the exposure is not so high as to destroy the relatively radioresistant small
primordial follicles. The dose that will produce permanent sterility is age dependent,
with higher doses (~10 Gy) required to produce sterility prior to puberty than in
premenopausal women over 40 years old (~2 to 3 Gy).
Another concern regarding gonadal irradiation is the induction of genetic mutations
and their effect on future generations. This subject is addressed later in the chapter.
Ocular Effects
The lens of the eye contains a population of radiosensitive cells that can be damaged
or destroyed by radiation. Insofar as there is no removal system for these damaged
cells, they can accumulate to the point at which they cause vision impairing cataracts. A unique aspect of cataract formation is that, unlike senile cataracts, which
typically develop in the anterior pole of the lens, radiation-induced cataracts are
781
TABLE 20-4PHYSICAL AND BIOLOGICAL MODIFIERS OF RADIATION-
INDUCED SKIN DAMAGE

FACTOR
EXAMPLES
COMMENT
Location of
irradiated skin
Relative radiosensitivity:
anterior aspect of the neck
. flexor surfaces of the
extremities . trunk > back
. extensor surfaces of
extremities . nape of the
neck . scalp > palms of the
hands > soles of feet
See Figure 20-18C demonstrating focal

scalp epilation
Size of exposed area
Smaller lesions heal faster

due to cell migration from
skin margin surrounding
the exposure area thus
accelerating wound closure
Benefit only significant for relatively

small lesions. Not typically a factor for
medical exposures where field sizes
are larger.
Dose Fractionation
Dry Desquamation Threshold:

Single expsoure ~14 Gy
3 Fractionations in 3 d ~27 Gy
Repair of sublethal damage to DNA

is completed within ~24 h, however
repopulation can take days, weeks or
even months to complete.
Patient-related
Increased radiosensitivity
examples: smoking,
poor nutritional status,
compromised skin integrity,
light colored skin, obesity,
DNA repair defects, prior
irradiation on the same area,
UV exposure
DNA repair defect examples: ataxia

telangiectasia, Fanconi anemia,
Bloom syndrome and xeroderma
pigmentosum. Other diseases, e.g.,
scleroderma, hyperthyroidism,
diabetes mellitus.
Patients are more prone to sun burns
and should minimize sun exposure
following radiation-induced skin injury.
Drugs
Some drugs are known to

increase radiosensitivity, e.g.,
actinomycin D, doxorubicin,
bleomycin, 5-fluorouracil
and methotrexate
Some chemotherapeutic agents,

(e.g., doxorubicin, etoposide,
paclitaxel, epirubicin), antibiotics (e.g.,
cefotetan), statins (e.g., simvastatin),
and herbal preparations can produce
an inflammatory skin reaction at the
site of prior irradiation (radiation
recall) weeks to years after exposure
at the same location.
Source: Adapted from information provided in Balter S, Hopewell JW, Miller DL, et al. fluoroscopically guided
interventional procedures: a review of radiation effects on patients skin and hair. Radiology 2010;254(2):326341.
caused by abnormal differentiation of damaged epithelial cells that begin as a small

opacities (abnormal lens fibers) in the anterior subcapsular region and migrate posteriorly. Even at relatively minor levels of visual acuity loss, these posterior subcapsular cataracts can impair vision by causing glare or halos around lights at night.
While the degree of the opacity and the probability of its occurrence increases with
the dose, the latent period is inversely related to dose. High-LET radiation is more
efficient for cataractogenesis by a factor of 2 or more. There have been a number of recent studies of mechanistic models of radiation-induced cataractogenesis.
In addition, more recent epidemiological studies have included several additional
occupational exposure populations and longer periods of observation for previously
studied populations. These studies have raised concern regarding the previous scientific consensus that regarded radiation-induced cataracts as a deterministic effect
782
with dose thresholds for detectable opacities of 2 for acute and 5 Gy for chronic
exposures respectively. The view that cataractogenesis is a deterministic event,
exhibiting a dose threshold below which lens opacities would not develop, served
as the basis for ICRP and NCRP recommended occupational dose limit to the lens
of the eye of 150 mSv/y (ICRP, 1991, 2007; NCRP, 2000). Studies of A-bomb survivors who were young at the time of exposure and followed for longer periods than
previous studies and other exposed populations such as the workers involved in
the cleanup around the Chernobyl nuclear reactor accident site and radiologic technologists in the United States (Gabriel, 2008), suggest that if there is a threshold for
cataract development, it is likely to be substantially lower than previously believed
(Ainsbury, 2009; ICRP, 2011). These data, and the presumption that subclinical, but
detectable opacities will, if given enough time, eventually progress to impair vision
led the ICRP to conclude in a recent statement that the threshold for acute and
chronic exposure may be more on the order of 0.5 Gy (ICRP, 2011a). Furthermore,
some suggest that the dose response may be more accurately described by a linear
no-threshold stochastic (rather than a deterministic) model. ICRPs recent review of
the scientific evidence regarding the risk of radiation-induced cataract has led the
commission to propose a much more conservative occupational equivalent dose
limit for the lens of the eye (20 mSv/y averaged over 5 years, with no single year
exceeding 50 mSv).
Cataracts among early radiation workers were common because of the extremely
high doses resulting from long and frequent exposures from poorly shielded x-ray
equipment and the absence of any substantial shielding of the eyes. Today, radiationinduced cataracts are much less common; however, there is concern that for radiation
workers receiving higher lens exposures in a medical setting (typically from interventional fluoroscopic procedures) there may be a risk for clinically significant lens
opacities over an occupational lifetime. Considering the mounting evidence of a substantially lower threshold for radiation-induced cataracts, the current US regulatory
limit of 150 mSv/y to the lens of the eye may need to be reevaluated. However, the
proposed ICRP limit is almost a factor of 10 lower than current limits and lower
than the whole body dose limit in the United States of 50 mSv/y. Adoption of ICRP
recommendations by regulatory bodies would present new challenges for radiation
protection in health care settings, especially for those involved in performing fluoroscopically guided interventional procedures. In any case, the use of eye protection in
the form of leaded glasses or ceiling mounted lead acrylic shielding is imperative for
workers whose careers will involve long-term exposure to scattered radiation.
Summary
There is general agreement that acute doses below 100 mGy will not result in any
functional impairment of tissues or organ systems. This can also be considered generally applicable to the risk of clinically significant lenticular opacities with the caveat
that the existence of a true deterministic threshold for radiation-induced cataracts
remains uncertain. The previous discussion has been limited to tissues and organ
systems that are often the focus of concerns for patients and for staff performing
diagnostic and interventional fluoroscopic procedures. A more complete discussion
of these and other organ and tissue reactions to radiation exposure can be found in
the current draft of the ICRP report devoted to this subject (ICRP, 2011). A summary
from this report of threshold doses (defined as ~1% incidence in morbidity) in tissues and organs in adults exposed to acute, fractionated or protracted, and chronic
irradiation is reproduced in Table 20-5.
783
TABLE 20-5THRESHOLD DOSES IN TISSUES AND ORGANS IN ADULTS
EXPOSED TO ACUTE, FRACTIONATED OR PROTRACTED, AND

CHRONIC RADIATION EXPOSUREa
EFFECT
ORGAN/TISSUE
HIGHLY
FRACTIONATED
(2 Gy PER
FRACTION) OR
EQUIVALENT
TIME TO ACUTE
PROTRACTED
DEVELOP EXPOSURE EXPOSURES
EFFECT
(Gy)
(Gy)b
Temporary sterility
Testes
39 wk
~0.1
NA
0.4
Permanent sterility
Testes
3 wk
~6
,6
2.0
Permanent sterility
Ovaries
,1 wk
~3
6.0
.0.2
Depression of
hematopoiesis
Bone marrow
37 d
~0.5
~1014 Gy
.0.4
Xerostomia
Salivary glands 1 wk
NA
,20
NA
Dysphasia, stricture
Esophagus
38 mo
NA
55
NA
Dyspepsia, ulceration
Stomach
2y
NA
50
NA
Stricture
Small intestine
1.5 y
NA
45
NA
Stricture
Colon
2y
NA
45
NA
Anorectal dysfunction
Rectum
1y
NA
60
NA
Hepatomegaly, ascites Liver
2 wk to
3 mo
NA
,3032
NA
Main phase of skin

reddening
Skin (large
areas)
14 wk
,36
30
NA
Skin burns
Skin (large
areas)
23 wk
510
35
NA
Temporary hair loss
Skin
23 wk
~4
NA
NA
Late atrophy
Skin (large
areas)
.1 y
10
40
NA
Telangiectasia @ 5 y
Skin (large
areas)
.1 y
10
40
NA
Cataract (visual
impairment)
Eye
.20 y
~0.5
~0.5
~0.5 divided
by years of
durationc
Acute pneumonitis
Lung
13 mo
67
18
NA
Edema
Larynx
45 mo
NA
70
NA
Renal failure
Kidney
.1 y
78
18
NA
Fibrosis/necrosis
Bladder
.6 mo
15
55
NA
Stricture
Ureters
.6 mo
NA
5560
NA
Fracture
Adult bone
.1 y
NA
50
NA
Fracture
Growing bone
,1 y
NA
25
NA
Necrosis
Skeletal
Muscle
Several
years
NA
55
NA
.10 y
NA
.18
NA
Endocrine dysfunction Thyroid
ANNUAL
(CHRONIC) DOSE
RATE FOR MANY
YEARS (Gy y1)
(Continued)
784
TABLE 20-5THRESHOLD DOSES IN TISSUES AND ORGANS IN ADULTS
EXPOSED TO ACUTE, FRACTIONATED OR PROTRACTED, AND

CHRONIC RADIATION EXPOSUREa (continued)
EFFECT
ORGAN/TISSUE
HIGHLY
FRACTIONATED
(2 Gy PER
FRACTION) OR
EQUIVALENT
TIME TO ACUTE
PROTRACTED
DEVELOP EXPOSURE EXPOSURES
EFFECT
(Gy)
(Gy)b
ANNUAL
(CHRONIC) DOSE
RATE FOR MANY
YEARS (Gy y1)
Endocrine dysfunction Pituitary
.10 y
NA
10
NA
Paralysis
Spinal cord
.6 mo
NA
55
NA
Necrosis
Brain
.1 y
NA
5560
NA
Cognitive defects
Brain
Several
years
12
,20
NA
Cognitive defects
infants ,18 mo
Brain
Several
years
0.10.2
NA
NA
Defined as 1% incidence in morbidity. Most values rounded to nearest Gy; ranges indicate area dependence
for skin and differing medical support for bone marrow.
b
Derived from fractionated radiotherapeutic exposures, generally using 2 Gy per fraction. For other fraction
sizes, the following formula can be used, where D is total dose (number of fractions multiplied by d), d is dose
per fraction (2 Gy in the case of D1, and new value of d in the case of D2), and the ratio a/b can be found in
the appropriate section of the ICRP (2011) report: D1[1+2/(a/b)] = D2[1+d2/(a/b)].
Note: Protracted doses at a low dose rate of around 10 mGy per minute are approximately isoeffective to
doses delivered in 2 Gy fractions at high dose rate for some tissues, but this equivalence is dependent on the
repair half-time of the particular tissue. Further details can be found in ICRP (2011) report references Joiner
and Bentzen, 2009; Bentzen and Joiner, 2009; van der Kogel, 2009.
c
The values quoted for the lens assume the same incidence of injury irrespective of the acute or chronic nature
of the exposure, with more than 20 years follow up. It is emphasized that great uncertain is attached to these
values.
NA, Not Available.
a
20.5 Whole Body Response to Radiation: The Acute
Radiation Syndrome
As previously discussed, the body consists of cells of differing radiosensitivities and

a large radiation dose delivered acutely yields greater cellular damage than the same
dose delivered over a protracted period. When the whole body (or large portion of
the body) is subjected to a high acute radiation dose, there are a series of characteristic clinical responses known collectively as the acute radiation syndrome (ARS).
The ARS is an organismal response quite distinct from isolated local radiation injuries such as epilation or skin ulcerations.
The ARS refers to a group of subsyndromes occurring in stages over a period
of hours to weeks after the exposure as the injury to various tissues and organ
systems is expressed. These subsyndromes result from the differing radiosensitivities of these organ systems. In order of their occurrence with increasing radiation
dose, the ARS is divided into the hematopoietic, gastrointestinal, and neurovascular
syndromes. These syndromes identify the organ system, the damage to which is,
primarily responsible for the clinical manifestation of disease. The ARS can occur
when a high radiation dose is (1) delivered acutely; (2) involves exposure to the
785
whole body (or at least a large portion of it), and (3) is from external penetrating
radiation, such as x-rays, gamma rays, or neutrons. Accidental internal or external
contamination with radioactive material is unlikely to result in a sufficiently acute
dose to produce the ARS in the organ systems. However, as the widely publicized
death of Alexander Litvinenko in 2006 from Po-210 (an alpha emitter) poisoning demonstrated, ARS can be observed when internal contamination with large
quantities of highly radiotoxic material (~2 GBq in this case) are widely distributed in the body. Mr. Litvinenko died approximately 3 weeks after the poisoning
from the complications of profound panocytopenia that is characteristic of severe
hematopoietic damage.
Sequence of Events
The clinical manifestation of each of the subsyndromes occurs in a predictable
sequence of events that includes the prodromal, latent, manifest illness, and, if the dose
is not fatal, recovery stages (Fig. 20-19).
The onset of prodromal symptoms is dose dependent and can begin within
minutes to hours after the exposure. As the whole-body exposure increases above a
threshold of approximately 0.5 to 1 Gy, the prodromal symptoms, which (depending
on dose) can include anorexia, nausea, lethargy, fever, vomiting, headache, diarrhea
and altered mental status, begin earlier and are more severe. Table 20-6 summarizes
Irradiation
0-48 hours
hours - 21 d
hours - 30 d
PRODROMAL
LATENT
MANIFEST
ILLNESS
Anorexia
Nausea
Vomiting
Diarrhea
Fever
Lymphopenia
Granulocytosis
Prodromata
absent or
diminished
hours
60+d
DEATH
RECOVERY?
Vascular damage
Infection
Bone marrow depression
Leukopenia
Thrombocytopenia
GI symptoms
As Dose Severity of Time to Onset of

Increases
Effects
Manifest Illness
Increase
Decreases
FIGURE 20-19 ARS follows a clinical pattern that can be divided into three phases: (1) an initial or prodromal phase that presents as nonspecific clinical symptoms, such as nausea, vomiting, and lethargy (hematological changes may also occur during this period); (2) the latent phase, during which the prodromal symptoms
typically subside; and (3) the manifest illness phase, during which the underlying organ system damage is
expressed. The type, time of onset, and severity of prodromal symptoms are dose dependent. The duration of
the latent period, as well as the time of onset and severity of the manifest illness phase, and ultimate outcome
are all, to a variable extent, dependent upon total dose, uniformity of the exposure, and individual radiation
sensitivity. As a rule, higher doses shorten the time of onset and duration of all three phases and increase the
severity of the prodrome and the manifest illness phases.
786
TABLE 20-6 CLINICAL FINDINGS DURING PRODROMAL PHASE OF ARS

ARS DEGREE AND THE APPROXIMATE DOSE OF ACUTE WHOLE BODY EXPOSURE
SYMPTOMS
AND MEDICAL
RESPONSE
MILD
(12 Gy)
MODERATE
(24 Gy)
SEVERE
(46 Gy)
VERY SEVERE
(68 Gy)
LETHAL (.8 Gy)a
Vomiting Onset 2 h after

12 h after
exposure or exposure
later
Earlier than 1 h Earlier than Earlier than 10 min

after exposure 30 min after after exposure
exposure
Incidence, %
1050
7090
100
100
100
Diarrhea
None
None
Mild
Heavy
Heavy
Onset
38 h
13 h
Within minutes
or 1 h
Incidence, %
,10
.10
Almost 100
Moderate
Severe
Severe
424 h
34 h
12 h
50
80
8090
Unaffected
May be
altered
Unconsciousness
(may last seconds
to minutes)
Headache
Slight
Mild
Onset
Incidence, %
Consciousness
Unaffected
Unaffected
Onset
Seconds/minutes
Incidence, %
100 (at ,50 Gy)
Body
temperature
Increased
Fever
High fever
High fever
Onset
13 h
12 h
,1 h
,1 h
Incidence, %
1080
80100
100
100
Medical
response
Normal
Outpatient Observation Treatment in

observation in general
specialized
hospital,
hospital
treatment in
specialized
hospital if
needed
Treatment in Palliative
specialized
treatment
hospital
(symptomatic only)
ARS, Acute radiation syndrome;

a
With intensive medical support and marrow resuscitative therapy, individuals may survive for 6 to 12 months
with whole-body doses as high as 12 Gy.
Source: Adapted from Diagnosis and Treatment of Radiation Injuries, Safety Report Series No. 2. Vienna,
Austria: International Atomic Energy Agency, 1998 and Koenig KL, Goans RE, Hatchett RJ, et al. Medical
treatment of radiological casualties: current concepts. Ann Emerg Med 2005;45:643652.
some of the clinical findings, probability of occurrence, and time of onset that may be
anticipated during the prodromal phase of ARS as a function of whole body dose.
The time of onset and the severity of these symptoms were used during the
initial phases of the medical response to the Chernobyl (Ukraine) nuclear reactor
accident in 1986 to triage patients with respect to their radiation exposures. The
prodromal symptoms subside during the latent period, whose duration is shorter
for higher doses and may last for up to 4 weeks for modest exposures less than 1Gy.
The latent period can be thought of as an incubation period during which the
organ system damage is progressing. The latent period ends with the onset of the
clinical expression of organ system damage, called the manifest illness stage, which
787
can last for approximately 2 to 4 weeks or in some cases even longer. This stage is
the most difficult to manage from a therapeutic standpoint, because of the overlying immunoincompetence that results from damage to the hematopoietic system.
Therefore, treatment during the first 6 to 8 weeks after the exposure is essential
to optimize the chances for recovery. If the patient survives the manifest illness
stage, recovery is likely; however, the patient will be at higher risk for cancer and,
to a lesser extent, his or her future progeny may have an increased risk of genetic
abnormalities.
Hematopoietic Syndrome
Hematopoietic stem cells are very radiosensitive. However, with the exception of
lymphocytes, their mature counterparts in circulation are relatively radioresistant.
Hematopoietic tissues are located at various anatomic sites throughout the body;
however, posterior radiation exposure maximizes damage because the majority of
the active bone marrow is located in the spine and posterior region of the ribs and
pelvis. The hematopoietic syndrome is the primary acute clinical consequence of
an acute radiation dose between 0.5 and 10 Gy. Healthy adults with proper medical care almost always recover from doses lower than 2 Gy, whereas doses greater
than 8 Gy are almost always fatal unless advanced therapies such as the use of
colony-stimulating factors or bone marrow transplantation are successful. Growth
factors such as granulocyte-macrophage colony-stimulating factor and other glycoproteins that induce bone marrow hematopoietic progenitor cells to proliferate and
differentiate into specific mature blood cells have shown promise in the treatment
of severe stem cell depletion. Even with effective stem cells therapy, however, it is
unlikely that patients will survive doses in excess of 12 Gy because of irreversible
damage to the gastrointestinal tract and the vasculature. In the absence of medical
care, the human LD50/60 (the dose that would be expected to kill 50% of an exposed
population within 60 days) is approximately 3.5 to 4.5 Gy to the bone marrow. The
LD50/60 may extend to 5 to 6 Gy with supportive care such as the use of transfusions
and antibiotics and may be as high as 6 to 8 Gy with effective use of hematopoietic growth factors in an intensive care setting. In contrast to whole body highdose penetrating radiation exposures, radiation exposure during some accident
scenarios may result in nonuniform or inhomogeneous exposures for which the
potential for spontaneous hematopoietic regeneration from unirradiated or only
mildly irradiated stem cells is much greater. The probability of recovering from
a large radiation dose is reduced in patients who are compromised by trauma or
other serious comorbidities. The severe burns and trauma received by some of the
workers exposed during the Chernobyl nuclear accident resulted in a lower LD50/60
than would have been predicted from their radiation exposures alone. In addition,
patients with certain inherited diseases that compromise DNA repair, such as A-T,
Fanconis anemia, and Blooms syndrome, are known to have an increased sensitivity to radiation exposure.
The prodromal symptoms associated with the hematopoietic syndrome can occur
within a few hours after exposure and may consist of nausea, vomiting, headache,
and diarrhea. If these symptoms appear early and severe diarrhea occurs within the
first 2 days, the radiation exposure may prove to be fatal. The prodromal and latent
periods may each last for weeks. Although the nausea and vomiting may subside during the latent period, patients may still feel fatigued and weak. During this period,
damage to the stem cells reduces their number and thus their ability to maintain
normal hematologic profiles by replacing the circulating blood cells that eventually
788
Irradiated Marrow
Normal Marrow
FIGURE 20-20 A. Normal bone marrow stem cells. B. Pyknotic stem cell damage following a bone marrow dose of approximately 2 Gy. (Adapted from Medical management of radiological casualties, Online Third
Edition June 2010 Armed Forces Radiobiology Research Institute, Bethesda, MD http://www.usuhs.mil/afrri/
outreach/pdf/3edmmrchandbook.pdf)
die by senescence. The kinetics of this generalized pancytopenia are accelerated with
higher (acute) exposures. An example of radiation-induced stem cell damage following a bone marrow dose of 2 Gy is shown in Figure 20-20. Figure 20-21 illustrates the
time course of the hematological consequences of bone marrow doses of 1 and 3 Gy.
14
Thrombocytes
12
Hemoglobin
10
Hemoglobin (g)
12
200
Neutrophils
100
Lymphocytes
10
20
30
40
50
60
14
12
Thrombocytes
Hemoglobin (g)
10
300
Hemoglobin
16
12
2
0
Thrombocytes (x 1000)
Lymphocytes and
Neutrophils (x 1000)
300
16
200
Lymphocytes
100
Neutrophils
0
10
20
30
40
50
60
Time (days)
FIGURE 20-21 Hematological changes following an acute bone marrow dose of 1 Gy (A) and 3 Gy (B). One
Gy causes a transient and 3 Gy an extended period of neutropenia, and thrombocytopenia. The lymphopenia is a consequence of radiation-induced apoptosis in some types of lymphocytes. (Adapted from Medical
management of radiological casualties. Online Third Edition June 2010 Armed Forces Radiobiology Research
Institute, Bethesda, MD http://www.usuhs.mil/afrri/outreach/pdf/3edmmrchandbook.pdf)
789
The initial rise in the neutrophil count is presumably a stress response in which
neutrophils are released from extravascular stores. The decline in the lymphocyte
count occurs within hours after exposure and is a crude early biologic marker of the
magnitude of exposure. The threshold for a measurable depression in the blood lymphocyte count is approximately 0.25 Gy; an absolute lymphocyte count lower than
1,000/mm3 in the first 48 hours indicates a severe exposure.
The clinical manifestation of bone marrow depletion peak 3 to 4 weeks after
the exposure as the number of cells in circulation reaches its nadir. Hemorrhage
from platelet loss and opportunistic infections secondary to severe neutropenia are
the potentially lethal consequences of severe hematopoietic compromise. Overall,
the systemic effects that can occur from the hematopoietic syndrome include mild
to profound immunologic compromise, sepsis, hemorrhage, anemia, and impaired
wound healing.
Gastrointestinal Syndrome
At higher doses the clinical expression of the gastrointestinal syndrome becomes
the dominant component of the radiation response, the consequences of which are
more immediate, severe, and overlap with those of the hematopoietic syndrome. At
doses greater than 12 Gy, this syndrome is primarily responsible for lethality. Its prodromal stage includes severe nausea, vomiting, watery diarrhea, and cramps occurring within hours after the exposure, followed by a much shorter latent period (5 to
7 days). The manifest illness stage begins with the return of the prodromal symptoms that are often more intense than during their initial presentation. The intestinal
dysfunction is the result of the severe damage to the intestinal mucosa. Severely
damaged crypt stem cells lose their reproductive capacity. As the mucosal lining
ages and eventually sloughs, the differentiated cells in the villi are not adequately
replaced by cells from the progenitor compartment in the crypt. The denuding of
bowel villi, in turn, causes a host of pathophysiological sequelae. The breakdown of
the mucosal barrier allows for the entry of luminal contents such as antigens, bacterial products, and digestive enzymes into the intestinal wall ultimately resulting
in a radiation-induced intestinal mucositis. The net result is a greatly diminished
capacity to regulate the absorption of electrolytes and nutrients and, at the same
time, a portal is created for intestinal flora to enter the systemic circulation. These
changes in the gastrointestinal tract are compounded by equally drastic changes in
the bone marrow. The most potentially serious effect is the severe decrease in circulating white cells at a time when bacteria are invading the bloodstream from the
gastrointestinal tract.
Overall, intestinal pathology includes mucosal ulceration and hemorrhage, disruption of normal absorption and secretion, alteration of enteric flora, depletion of
gut lymphoid tissue, and disturbance of gut motility. The systemic effects of acute
radiation enteropathy include malnutrition resulting from malabsorption; vomiting
and abdominal distention from paralytic ileus; anemia from gastrointestinal bleeding; sepsis resulting from invasion of intestinal bacteria into the systemic circulation;
and dehydration and acute renal failure from fluid and electrolyte imbalance. The
patient may not become profoundly panocytopenic, because death will likely occur
before radiation-induced damage to the bone marrow causes a significant decrease
in cell types with longer life spans (e.g., platelets and red cells). Lethality from the
gastrointestinal syndrome is essentially 100%. Death occurs within 3 to 10 days after
the exposure if no medical care is given or as long as 2 weeks afterward with intensive medical support.
790
It is important to appreciate that even at doses within the hematopoietic syndrome dose range (2 to 10 Gy), damage to the gastrointestinal tract is occurring. It
is responsible for many of the prodromal symptoms and contributes to the toxicity
of the radiation-induced myelosuppression that is the signature of the hematopoietic component of the ARS. While a whole body dose of 6 Gy does not result in the
full gastrointestinal sequelae described above, damage to the mucosal barrier causes
cytokines and other inflammatory mediators to be released into the circulation. In
addition, sepsis resulting from the entry of bacteria from the bowel into the systemic
circulation during a period of progressive neutropenia is an important cause of death
from doses in the hematopoietic syndrome dose range.
Neurovascular Syndrome
Death occurs within 2 to 3 days after supralethal doses in excess of 50 Gy. Doses in
this range result in cardiovascular shock with a massive loss of serum and electrolytes
into extravascular tissues. The ensuing circulatory problems of edema, increased
intracranial pressure, and cerebral anoxia cause death before damage to other organ
systems and tissues can become clinically significant.
The stages of the neurovascular syndrome are extremely compressed. Patients
may experience transitory incapacitation or unconsciousness. The prodromal
TABLE 20-7CLINICAL FEATURES DURING THE MANIFEST ILLNESS PHASE
OF ARS
DEGREE OF ARS AND APPROXIMATE DOSE OF ACUTE WHOLE-BODY EXPOSURE
MILD
(12 Gy)
Onset of signs
MODERATE
(24 Gy)
SEVERE
(46 Gy)
VERY SEVERE
(68 Gy)
LETHAL
(.8 Gy)
.30 d
1828 d
818 d
,7 d
,3 d
Lymphocytes, G/L
0.81.5
0.50.8
0.30.5
0.10.3
0.00.1
Platelets, G/La
60100
3060
2535
1525
,20
Percent of patients 10%25%

with cytopenia
25%40%
40%80%
60%80%
80%100%b
Clinical
manifestations
Fatigue,
weakness
Fever,
infections,
bleeding,
weakness,
epilation
High fever,
infections,
bleeding,
epilation
High fever,
diarrhea,
vomiting,
dizziness and
disorientation,
hypotension
High fever,
diarrhea,
unconsciousness
Lethality, %
050
2070
50100
100
68 wk
48 wk
12 wk
12 wk
Special
prophylactic
treatment
from d 1420;
isolation from
d 1020
Special
prophylactic
treatment
from d 710;
isolation from
the beginning
Special
12 wk
treatment from Symptomatic
the first day;
only
isolation from
the beginning
Onset
Medical response
Prophylactic
G/L, SI units for concentration and refers to 109 per liter.

In very severe cases, with a dose greater than 50 Gy, death precedes cytopenia.
Source: Adapted from Diagnosis and Treatment of Radiation Injuries. Safety Report Series No. 2. Vienna,
Austria: International Atomic Energy Agency, 1998 and Koenig KL, Goans RE, Hatchett RJ, et al. Medical treatment of radiological casualties: current concepts. Ann Emerg Med 2005;45:643652.
791
period may include a burning sensation of the skin that occurs within minutes,
followed by nausea, vomiting, confusion, ataxia, and disorientation within 1 hour.
There is an abbreviated latent period (4 to 6 hours), during which some improvement is noted, followed by a severe manifest illness stage. The prodromal symptoms return with even greater severity, coupled with respiratory distress and gross
neurologic changes (including tremors and convulsions) that inevitably lead to
coma and death. Many other aspects of this syndrome are not understood because
human exposures to supralethal radiation are rare. Experimental evidence suggests
that the initial hypotension may be caused by a massive release of histamine from
mast cells, and the principal pathology may result from massive damage to the
microcirculation.
Summary of Clinical Features During the Manifest Illness

Phase of the Acute Radiation Syndrome
Table 20-7 summarizes the clinical features of the ARS within several dose ranges
with respect to hematological changes, manifestations of clinical symptoms,
latency as well as the medical response, and probability of survival. The previously discussed relationships among various elements of the ARS are summarized
in F
igure20-22.
Gastrointestinal System
food intake
absorption
diarrhea
ulceration
Poor Nutrition
Fluid Loss
Electrolyte Loss
Hematopoietic System
lymphocytes
granulocytes
Infection
platelets
erythrocytes
Hemorrhage
Vascular System
capillary permeability
vascular fragility
Anemia
obstruction of vessels
Anoxia
Damage to More
Resistant Tissues
FIGURE 20-22 Relationships among various elements of the ARS.
792
20.6 Radiation-Induced Carcinogenesis
Introduction
Most of the radiation-induced biologic effects discussed thus far are detectable within
a relatively short time after the exposure. Ionizing radiation can, however, cause damage whose expression is delayed for years or decades. The ability of ionizing radiation
to increase the risk of cancer years after exposure has been well established. Cancer,
unfortunately, is not a rare disease; indeed, it is the second most likely cause of death,
after cardiovascular disease, in the United States. According to a recent report on
cancer in the U.S. from the American Cancer Society (ACS, 2011) the lifetime probability of developing an invasive cancer is 41% (44.3% male and 37.8% female) and
the probability of dying from cancer is about half that, 22% (23.2% males and 19.7%
female). Incidence rates are defined as the number of people per 100,000 who are
diagnosed with cancer during a given time period (typically provided as an annual
age-adjusted rate per 100,000). The National Center for Health Statistics recommends
that the U.S. 2000 standard population be used when calculating and comparing ageadjusted rates for specific population groups. If one compares age-adjusted rates from
different populations, it is important that the same standard population is used for the
comparison. The annual cancer incidence and mortality age-adjusted rates, for the U.S.
population, are approximately 465 and 178 per 100,000, respectively, with males
having a higher incidence rate (543 per 100,000) than females (409 per 100,000),
(CDC, 2011). Subsequent sections of this chapter address estimates of radiationinduced cancer (incidence and mortality). To help place these numbers in perspective, the baseline lifetime risk of cancer incidence, mortality and average years of
life lost are shown in Table 20-8. As previously discussed, ICRP tissue weighting
factors (wT) were developed to account for inherent differences in tissue sensitivity to
detriment caused by radiation exposure. Part of the detriment values are based on
average years of life lost, which, even in the absence of additional radiation exposure,
can vary between sex and by type of cancer by almost a factor of two (e.g., prostate
and breast cancer).
Although the etiologies of most cancers are not well defined, diet, lifestyle,
genetic, and environmental conditions appear to be among the most important factors affecting specific cancer risks. For example, the total cancer incidence among
populations around the world varies by only a factor of 2 or so, but the incidences of
specific cancers can vary by a factor of 200 or more!
Cancer is the most important delayed somatic effect of radiation exposure. However,
radiation is a relatively weak carcinogen at low doses (e.g., occupational and diagnostic
exposures). While moderate doses of radiation cause well-documented effects, one
cannot detect significantly increased effects at the doses typically encountered in diagnostic imaging. In fact, the bodys robust capacity to repair radiation damage means
that the possibility of no increased risk at low doses cannot be ruled out. The effectiveness of different DNA repair systems was discussed above. The determinants of
radiation-induced cancer risk are discussed in greater detail later in the chapter.
Molecular Biology and Cancer

Cancer arises from abnormal cell division. Cells in a tumor are believed to descend
from a common ancestral cell that at some point (typically decades before a tumor
results in clinically noticeable symptoms) loses its control over normal reproduction.
The malignant transformation of such a cell can occur through the accumulation of
793
TABLE 20-8BASELINE LIFETIME RISK ESTIMATES OF
CANCER INCIDENCE AND MORTALITY

INCIDENCE
MORTALITY
CANCER SITE
MALES
FEMALES
Solid cancera
45,500
36,900
MALES
FEMALES
22,100 (11)
17,500 (11)
Stomach
1,200
720
670 (11)
430 (12)
Colon
4,200
4,200
2,200 (11)
2,100 (11)
Liver
640
280
490 (13)
260 (12)
Lung
7,700
5,400
7,700 (12)
4,600 (14)
Breast
12,000
3,000 (15)
15,900
3,500 (8)
Uterus
3,000
750 (15)
Ovary
1,500
980 (14)
Prostate
Bladder
3,400
1,100
770 (9)
330 (10)
12,500
8,800
6,800 (13)
5,100 (13)
Thyroid
230
550
40 (12)
60 (12)
Leukemia
830
590
710 (12)
530 (13)
Other solid cancer
Note: Number of estimated cancer cases or deaths in population of 100,000

(No. of years of life lost per death).
a
Solid cancer incidence estimates exclude thyroid and nonmelanoma skin
cancers.
Source: BEIR. Health risks from exposure to low levels of ionizing radiation: BEIR
VII, Phase 2. Committee to Assess Health Risks from Exposure to Low Levels of
Ionizing Radiation, Board of Radiation Effects, Research Division on Earth and
Life Studies, National Research Council of the National Academies. National
Academy of Sciences, Washington, DC: National Academies Press, 2006.
mutations in specific classes of genes. Mutations in these genes are a critical step in
the development of cancer.
Any protein involved in the control of cell division may also be involved in cancer. However, two classes of genes, tumor suppressor genes and protooncogenes, which
respectively inhibit and encourage cell growth, play major roles in triggering cancer.
Tumor suppressor genes such as the p53 gene, in its nonmutated or wild-type state,
promotes the expression of certain proteins. One of these halts the cell cycle and gives
the cell time to repair its DNA before dividing. Alternatively, if the damage cannot be
repaired, the protein pushes the cell into apoptosis. The loss of normal function of
the p53 gene product may compromise DNA repair mechanisms and lead to tumor
development. Defective p53 genes can cause abnormal cells to proliferate and as
many as 50% of all human tumors have been found to contain p53 mutations.
Protooncogenes code for proteins that stimulate cell division. Mutated forms of
these genes, called oncogenes, can cause the stimulatory proteins to be overactive, resulting in excessive cell proliferation. For example, mutations of the Ras protooncogenes
(H-Ras, N-Ras, and K-Ras) are found in about 25% of all human tumors. The Ras family of proteins plays a central role in the regulation of cell growth and the integration
of regulatory signals. These signals govern processes within the cell cycle and regulate
cellular proliferation. Most mutations result in abrogation of the normal enzymatic
activity of Ras, which causes a prolonged activation state and unregulated stimulation
of Ras signaling pathways that either stimulate cell growth or inhibit apoptosis.
794
Stages of Cancer Development

Cancer is thought to occur as a multistep process in which the initiation of damage
in a single cell leads to a preneoplastic stage followed by a sequence of events that
permit the cell to successfully proliferate. All neoplasms and their metastases are
thought to be derivatives or clones of a single cell and are characterized by unrestrained growth, irregular migration, and genetic diversity. For the purpose of setting
radiation protection standards, it is assumed that there is no threshold dose for the
induction of cancer, because even a single ionization event could theoretically lead to
molecular changes in the DNA that result in malignant transformation and ultimately
cancer. However, the probability of cancer development is far lower than would be
expected from the number of initiating events. For example, a whole body dose of
3 mGy of low-LET radiation (equivalent to the average annual background in the
United States) generates multiple DNA lesions (including on average, three SSBs and
five to eight damaged bases) in every cell (BEIR, 2006). However, cancer may never
arise because a host of defense mechanisms are initiated following radiation-induced
damage to prevent cancer development (e.g., activation of DNA repair systems; free
radial scavenging; cell cycle checkpoint controls; induced apoptosis, mitotic failure,
etc.). Additionally, all of the subsequent steps required for expression of the malignant potential of the cell may not occur.
Cancer formation can be thought of (albeit in a greatly oversimplified way) as occurring in three stages: (1) initiation, (2) promotion, and (3) progression. During initiation,
a somatic mutational event occurs that is misrepaired. This initial damage can be produced by radiation or any of a variety of other environmental or chemical carcinogens.
During the promotion stage, the preneoplastic cell is stimulated to divide. A promoter
is an agent that by itself does not cause cancer but, once an initiating carcinogenic event
has occurred, promotes or stimulates the cell containing the original damage. Unlike
many carcinogens, radiation may act as an initiator or a promoter. Some hormones act
as promoters by stimulating the growth of target tissues. For example, estrogen and
thyroid-stimulating hormone may act as promoters of breast cancer and thyroid cancer,
respectively. The final stage is progression, during which the transformed cell produces
a number of phenotypic clones, not all of which are neoplastic. Eventually, one phenotype acquires the selective advantage of evading the hosts defense mechanisms, thus
allowing the development of a tumor and possibly a metastatic cancer. Radiation may
also enhance progression by immunosuppression resulting from damage to lymphocytes and macrophages that are essential to the humoral antibody response.
Environmental Risk Factors

Environmental factors implicated in the promotion of cancer include tobacco, alcohol, diet, sexual behavior, air pollution, and bacterial and viral infections. Support for
the role of environmental factors comes from observations such as the increased incidence of colon and breast cancer among Japanese immigrants to the United States
compared with those living in Japan. Among the best known and striking modifiers
of cancer risk is smoking. For men, relative risk (RR) of developing lung cancer is
20 to 40 times greater in smokers than nonsmokers. In addition, agents that compromise the immune system, such as the human immunodeficiency virus, increase the
probability of successful progression of a preneoplastic cell into cancer. A number of
chemical agents, when given alone, are neither initiators nor promoters but when given
in the presence of an initiator will enhance cancer development. Many of these agents
are present in cigarette smoke, which may in part account for its potent carcinogenicity. Environmental exposure to nonionizing radiation, such as radiofrequency radiation
795
from cellular telephones and their base stations or magnetic field exposures from power
lines, has been an area of intense research in the last few decades. While reports of possible associations between sources of nonionizing radiation and cancer have received
considerable media attention, the evidence of any causal connection is weak and inconsistent and no biologically plausible mechanism of action has been identified.
A number of national and international agencies and organizations such as the World
Health Organizations International Agency for Research on Cancer (IARC) and the
National Toxicology Program (NTP) (an inter-agency program within the U.S. Department of Health and Human Services) report on the carcinogenic potential of various
physical and chemical agents. Each organization has its own rules and classification
schemes, which often lead to confusion in the public and the media regarding the potential health impact of a substance that appears in one of these reports. Periodically, the NTP
publishes an update of its Report on Carcinogens (RoC). This report, now in its 12th edition (NTP, 2011), identifies substances that are considered to have carcinogenic potential
in humans. The NTP lists more than 240 substances, which are classified into one of
two categories: known to be carcinogenic in humans, of which there are 54 in the current report including benzene, smoking tobacco, vinyl chloride, asbestos and of course,
ionizing radiation; and another 186 agents that are classified as reasonably anticipated
to be human carcinogens including exogenous progesterone used for contraception,
chemotherapeutic agents adriamycin and cisplatin as well as a naturally occurring contaminants such as aflatoxin, formed by certain fungi on agricultural crops.
Risk Expressions
One way of expressing the risk from radiation (or any other agent) in an exposed
population is in terms of its relative risk (RR). RR is the ratio of the disease (e.g., cancer) incidence in the exposed population to that in the general (unexposed) population; thus, a RR of 1.2 would indicate a 20% increase over the spontaneous rate that
would otherwise have been expected. The excess relative risk (ERR) is simply RR 1;
in this case, 1.2 1 0.2. While the RR of a specific cancer following some exposure
is informative about the magnitude of the increased risk relative to its natural occurrence, it does not provide a sense of perspective of the risk in terms of the overall health
impact the risk represents. For example, a study showing that a particular exposure
resulted in a 300% increase (i.e., RR of 4) in the incidence of a very rare cancer with a
natural incidence of 2 per 100,000 means that the cancer risk is now 8 per 100,000.
When compared to total incidence of cancer in the population of approximately
43,000 per 100,000, a 300% increase of a rare disease does not seem so impressive
as a potential health threat. Absolute risk (AR) is another way of expressing risk, as the
number of excess cancer cases per 100,000 in a population. In radiation epidemiology, it may be expressed as a rate such as the number of excess cases per 104 or 105
people per Sv per year (e.g., #/104/Sv/y). For a cancer with a radiation-induced risk of
4/104/Sv/y (or 4104 Sv1 y1) and a minimum latency period of about 10 years, the
risk of developing cancer within the next 40 years from a dose of 0.1 Sv would be
30 years 0.1 Sv 4 104 Sv1 y1 12 per 10,000 or 0.12%. In other words, if
10,000 people (with the same age and gender distribution as in the general population) each received a dose of 0.1 Sv, 12 additional cases of cancer would be expected
to develop in that population over the subsequent 40 years. Excess Absolute Risk
(EAR), also referred to as excess attributable risk, is the difference between two absolute risks and is commonly used in radiation epidemiology expressed as the EAR per
unit dose. Thus if the absolute risk in a population exposed to 1 Sv was 95 105 y1
and 20 105 y1 in the unexposed population, the EAR would be (95 per 100,000
per year) (20 per 100,000 per year) 75 105 y1 Sv1.
796
Modifiers of Radiation-Induced Cancer Risk

Radiation-induced cancers can occur in most tissues of the body and are indistinguishable from those that arise from other causes. The probability of development
of a radiation-induced cancer depends on a number of physical and biological factors. Physical factors include the radiation quality (e.g., LET), total dose and, in
some instances, the rate at which the dose was received (e.g., acute versus chronic).
Research on the biological factors that may influence the carcinogenic effectiveness
of radiation has been undertaken in numerous biomolecular, cellular, animal, and
epidemiological studies. The influence of radiation quality, dose fractionation, age,
tumor type, and gender on radiation-induced cancer risks following exposure and
the influence of genetic susceptibility to cancer will be discussed briefly below. Many
of these topics will be revisited in the context of specific results from major epidemiological investigations of radiation-induced cancer discussed in this chapter.
Radiation Quality
The RBE of radiation as a function of LET for a variety of biological end points (e.g.,
double strand DNA breaks, clonogenic potential of cells in culture) was discussed
earlier in this chapter. The high ionizations density of high-LET radiation is more
effective in producing DNA damage and is less likely to be faithfully repaired than
damage produced by low-LET radiation. Consequently, for a given absorbed dose, the
probability of inducing a cancer-causing mutation is higher for high-LET radiation,
but so is the probability of cell killing. Although the RBE values and their modifying
factors for radiocarcinogenesis are not known with great certainty, high-LET radiation
has been shown to produce more cancers of the lung, liver, thyroid, and bone than an
equal dose of low-LET radiation in human populations. The uncertainty in a-particle
risk is large, with a median value of 14.1 and a 90% CI from 5 to 40, (EPA, 2011). The
EPA, ICRP, and the NCRP recommend risk coefficients for a-particles that are based
on an RBE of 20. RBE values obtained from epidemiological studies vary greatly. For
example, studies of the patients injected with thorium-232 dioxide (a primordial alpha
emitting radionuclide with a half-life of billions of years) as a diagnostic contrast agent
(Thorotrast) for cerebral angiography from about 1930 to the mid 1950s, found an
RBE of approximately 20 for liver cancer, but an RBE of only about 1 for leukemia.
It has been demonstrated that for certain biological end points, such as efficiency of producing dicentrics in human lymphocytes, that low-energy photons are
more effective than high-energy photons (Fig. 20-23). Presumably this is a result of
the higher LET of lower energy secondary electrons (e.g., 30 keV electron LET ~ 1
keV/m) compared to that of higher energy electrons (e.g., 500 keV electron LET ~
0.2 keV/m) and the resultant increase in complex DNA damage generated by ionization and excitation events of these low-energy electrons near the ends of their tracks.
While there is experimental and theoretical evidence supporting higher RBEs of
low-photon and low-electron energies, epidemiological support for such an effect is
lacking. In fact, risk coefficients for x-rays derived from studies of medically irradiated cohorts are in some cases lower than what has been observed for the A-bomb
survivors. However, there are a number of potential confounders that may have prevented detection of an elevated risk if it were present. Thus any difference in carcinogenic risk, per unit dose, from low-energy x-rays compared to that of higher energy
photons remains to be determined (BEIR, 2006; ICRP, 2003a,b).
Dose Rate and Fractionation

It has long been recognized that the biological effectiveness of the radiation-induced
damage to cells and tissues generally decreases at lower dose rates. This effect is due
797

0.5
Dicentrics per cell / Gy
Human lymphocytes
0.1
137Cs
220kV
60Co
0.01
10
100
1000
Mean photon energy (keV)

FIGURE 20-23 The photon energy dependence for efficacy of producing dicentric chromosomes per unit dose
in human peripheral blood lymphocytes. Data points and standard errors of number of dicentrics per cell per Gy
for monoenergetic photons at low and high energies (x-axis) Squares are for monoenergetic photons; circles are
x-ray spectra or y-rays. (Source Sasaki MS, Kobayashi K, Hieda K, et al. Induction of chromosome aberrations in
human lymphocytes by monochromatic x-rays of quantum energy between 4.8 and 14.6 keV. Int J Radiat Biol
1989;56:975988; Sasaki MS. Primary damage and fixation of chromosomal DNA as probed by monochromatic
soft x-rays and low-energy neutrons. In: Fielden EM, ONeil P, eds. The Early Effects of Radiation on DNA. Vol. H54.
NATO ASI Series, Berlin: Springer-Verlag, 1991:369384; and BEIR VII.)
at least in part to the ability of cells to repair damage during exposure or between
exposures in the case of fractionated exposures. The effect of fractionating large doses
to increase the probability of cellular repair has been well characterized in vitro and
has been shown to reduce the incidence of carcinogenesis in some cases such as leukemia. Currently, a dose and dose-rate effectiveness factor (DDREF) are used to convert high-dose-rate risk estimates to estimates for exposure at low dose rates for the
purposes of radiation protection. However, at the low doses associated with diagnostic examinations and occupational exposures, dose rate may not affect cancer risk.
Age at Exposure, Gender, Tumor Type, and Latency

Latency (the period between exposure and clinical expression of disease) and the risk of
radiation-induced cancers vary with the type of cancer and age at the time of exposure. For
example, the risk of ovarian cancer from an acute exposure at age 10 is approximately three
times greater than if the exposure occurred at age 50. For whole body exposure, females on
average have a 40% higher risk of radiogenic cancer than males. This is due in large part
to the high risks for radiation-induced breast, ovarian, and lung cancer in women and the
substantially lower risks for radiation-induced testes and prostate cancer in men. Breast
cancer occurs almost exclusively in women, and absolute risk estimates for lung cancer
induction by radiation are (unlike the normal incidence) approximately twice as high for
women than for men. However, for some specific cancers, the radiation-induced cancer
risk is lower for women than for men (e.g., liver cancer ~50% lower risk in females than
males). The organs at greatest risk for radiogenic cancer induction and mortality are breast
and lung for women and lung and colon for men. The minimal latent period is 2 to 3 years
for leukemia, with a period of expression (i.e., the time interval required for full expression
of the radiogenic cancer increase) proportional to the age at the time of exposure, ranging
from approximately 12 to 25 years. Latent periods for solid tumors range from 5 to 40
years, with a period of expression for some cancers longer than 50 years.
798
Genetic Susceptibility
Mutations in one or more specific genes, while rare, are known to increase
susceptibility to developing cancer. Over the last few decades, extensive efforts have
been made to identify specific gene mutations that act as sources of genetic susceptibility to cancer. Increasing numbers of observational studies investigating the association between specific gene variants and cancer risk have been published. This effort
has been greatly accelerated by the mapping of the human genome and the results
from related advances aimed at identifying the quantity, type, location, and frequency
of genetic variants in human genes. This work, facilitated by advances in sequencing
technology, has allowed results to be obtained much faster and less expensively than
before and continues to contribute to the understanding of the genetic susceptibility
to cancer and to advance the goal of improved, individualized gene therapy.
While there are still many unanswered questions, the ability of specific inherited
gene mutations to substantially increase risk of developing specific cancers (i.e., high
penetrance genes) has been well documented. For example, women with inherited
mutations in the breast cancer susceptibility genes 1 or 2 (BRCA1 or BRCA2) and a
family history of multiple cases of breast cancer carry a lifetime risk of breast cancer
that is approximately 5 times higher than for women in the general population, (i.e.,
60% and 12%, respectively) (NCI 2011).
Ataxia-telangiectasia (A-T), a rare, recessive genetic disorder of childhood, occurs
in 12 of every 100,000 people. Patients with the ataxia-telangiectasia mutation (ATM)
have trouble walking as children (ataxia) and have small red spider-like veins (telangiectasia). These patients are at substantially higher risk of infection and of developing cancer (especially leukemias and lymphomas) than the general population. These
patients are also hypersensitive to ionizing radiation exposure because of defective
DNA repair mechanisms. The products of the ATM gene plays a central role in the
recognition and repair of double-strand DNA breaks and in the activation of cell cycle
checkpoints. A-T patients exhibit unusual susceptibility to injury by radiation and
often suffer more severe reactions to radiotherapy than other radiotherapy patients.
While physicians who treat A-T patients limit their x-ray exposures to the extent possible, they do recommend diagnostic x-ray imaging procedures when needed, if there
are not appropriate alternative procedures that do not use ionizing radiation.
There are still many open questions regarding these single gene human genetic disorders and their influence on cancer risks. One such question is to what extent radiation exposure modifies the cancer risk in patients with these inherited mutations. The
BEIR VII committee concluded that, while there is evidence to suggest that many of the
known, strongly expressing, cancer-prone human genetic disorders are likely to show
an elevated risk of radiation-induced cancer, the rarity of these disorders in the population will not significantly distort current population-based cancer risk estimates. Their
view was that the more practical issue associated with these high penetrance genes was
their impact on the risk of second cancers in such patients following radiotherapy.
Epidemiologic Investigations of Radiation-Induced Cancer

Although the dose-response relationship for cancer induction at high dose (and dose
rate) has been fairly well established for several cancers the same cannot be said for
low doses like those resulting from typical diagnostic and occupational exposures.
Insufficient data exist to determine accurately the risks of low-dose radiation exposure to humans. Animal and epidemiologic investigations indicate that the risks of
low-level exposure are small, but how small is still (despite decades of research) a
matter of great debate in the scientific community. Nevertheless, there is general
799
agreement that above cumulative doses of 100 to 150 mSv (acute or protracted exposure), direct epidemiological evidence from human populations demonstrates that
exposure to ionizing radiation increases the risk of some cancers.
The populations that form the bases of the epidemiologic investigation of radiation bioeffects come from four principal sources: (1) Life Span Study (LSS) cohort
of survivors of the atomic bomb explosions in Hiroshima and Nagasaki; (2) patients
with medical exposure during treatment of a variety of neoplastic and nonneoplastic
diseases; (3) persons with occupational exposures; and (4) populations with high
natural background exposures (Fig. 20-24).
It is very difficult to detect a small increase in the cancer rate due to radiation exposure at low doses (less than ~100 mSv) because radiation is a relatively weak carcinogen;
the natural incidence of many types of cancer is high and the latent period for most cancers is long. To rule out simple statistical fluctuations, a very large irradiated population
is required. To be able to detect a relative cancer risk of 1.2 with a statistical confidence of
95% (i.e., p , 0.05) when the spontaneous incidence is 2% in the population (typical of
many cancers), a study population in excess of 10,000 is required. More than 1 million
people would be required to identify a RR of 1.01 (i.e., a 1% cancer rate increase) in this
same population! A simplified hypothetical example that demonstrates the limited statistical power faced by many epidemiological studies of radiation exposure at low doses
was provided in an ICRP report on low-dose extrapolation of radiation-related cancer
risk (ICRP, 2006). Statistical power calculations were performed to assess the population
size needed for 80% power to detect an excess risk at the 5% significance level in which
baseline cancer risk, for an unspecified and hypothetical subset of cancers, is known to
be 10%, and the unknown radiation-related excess risk is actually 10% at 1 Gy and
proportional to dose between 0 and 1 Gy. As shown in Table 20-9, the population size
necessary to be able to detect an increased risk at doses typical of many diagnostic imaging exams (organ dose less than 10 mGy) would require a very large population and
enormous resources to accomplish. As pointed out by ICRP, the calculation is actually
unrealistically optimistic since, as one can never be that sure of the baseline rate in any
exposed population, it may be necessary to estimate the baseline rate by including an
equal number of nonexposed subjects (i.e., twice the population size would be required
to have equal power for detecting the difference). Confounding factors take on much
greater importance when excess risks are low, and spurious results can occur by chance
alone that result in exaggerated estimates of risk. Commenting on this problem, the ICRP
stated, At low and very low radiation doses, statistical and other variations in baseline
risk tend to be the dominant sources of error in both epidemiological and experimental
carcinogenesis studies, and estimates of radiation-related risk tend to be highly uncertain because of a weak signal-to-noise ratio and because it is difficult to recognize or
to control for subtle confounding factors. At such dose levels, and with the absence
of bias from uncontrolled variation in baseline rates, positive and negative estimates of
radiation-related risk tend to be almost equally likely on statistical grounds, even under
the LNT theory. Also, by definition, statistically significant positive or negative findings
can be expected in about one in 20 independent studies when the underlying true excess
risk is close to zero. Thus, even under the LNT theory, the smaller the dose, the more
likely it is that any statistically significant finding will be a purely chance occurrence, and
that it will be consistent with either beneficial effects of radiation (hormesis) or a grossly
exaggerated risk (Land, 1980).. A result predictable under both of two opposing
hypotheses supports neither of them against the other. Thus, for example, failure of
epidemiological studies to demonstrate a statistically significant excess cancer risk associated with exposures of the order of 1 mGy does not imply that there is no risk, although
it does suggest that any such risk is small relative to baseline cancer rates., ICRP 2006.
800
1918: It was believed that enlarged

Thymus gland in children was abnormal
and radiation was suggested to shrink it.
E
FIGURE 20-24 Sources of data on exposure of humans to radiation. The most important source of epidemiological data is the LSS of the Japanese atomic bomb survivors, who received acute doses of radiation,
over a range of doses up to 2 Gy, beyond which errors in dose reconstruction and mortality from complications of the ARS provided limited radioepidemiological information. The studies of cancer mortality in the
LSS began in 1950, and have formed the basis of radiation protection guidelines ever since. More recently,
cancer incidence data have been included by the ICRP in the formulation of radiation detriment, which is
used to establish tissue weighting factors for calculating effective dose in their radiation protection system. A. There was widespread destruction following the detonation of the atomic bomb in Hiroshima,
Japan at 8:15 AM on August 6, 1945. The building shown was the former Hiroshima Prefecture Industrial Promotion Hall, where special products of Hiroshima were exhibited and various gatherings were held.
(continued)
801
FIGURE 20-24. (Continued) Located just under the hypocenter, blast pressure was vertically exerted on the building and only the dome-shaped framework and part of the outer wall remained. B. From 1931 to the mid 1950s, Thorotrast, a colloidal suspention of radioactive thorium dioxide (top panel), was commonly used as a diagnostic contrast
agent for cerebral angiography. Thorotrast (containing thorium, a long lived alpha emitter) remains in the body, accumulates in the liver and results in liver cancer and leukemia. Thorotrast-laden macrophages in the bone marrow shown
at 1000 (bottom panel). (Graham et al., Whole-body pathologic analysis of a patient with thorotrast-in-induced
myelodysplasia. Health Phys 1992;63(1):2026.) C. Radiation has also been used in the past to treat benign medical
conditions with unfortunate consequences such as the increase in thyroid cancer in children who were unnecessarily irradiated to reduce the size of the thymus gland. During a review of children with thyroid cancer, Dr. Fitzgerald
noticed that nine out of the first ten patients he reviewed had had a history of thymic radiation. Drs. Duffy and
Fitzgeralds report, Cancer of the thyroid in children, in the Journal of Endocrinology in 1950 was the first demonstration of an association between radiation treatment and thyroid cancer. D. In the 1920s, bone cancer was
linked with ingestion of large quantities of radium by young women who painted dials on watches and clocks
with radium-laden paints. The type of bone cancer (osteogenic sarcoma) is rare, but it occurred with alarming
incidence in radium-dial painters and its location (often in the mandible) is an extremely unusual location for
this type of cancer. E. Several areas of the world have high natural background due to being at high elevation
or having high concentrations of naturally occurring radioactive material in the ground. High concentrations of
radioactive thorium-containing monazite sands are found in the coastal belt of Karunagappally, Kerala, India. The
median outdoor radiation levels are more than 4 mGy/y and, in some locations, as high as 70 mGy/y. A cohort of
all 385,103 residents in Karunagappally was established in the 1990s to evaluate the health effects of living in a
high background radiation area. Studies to date however have not shown any excess cancer risk from this chronic
exposure to gamma radiation (Nair RR, Rajan B, Akiba S, et al. Background radiation and cancer incidence in
Kerala, India-Karanagappally cohort study. Health Phys. 2009;96:5566; and Boice JD Jr, Hendry JH, Nakamura N,
et al. Low-dose-rate epidemiology of high background radiation areas. Radiat Res 2010;173:849854.)
Considering the limitations (both practical and inherent) to epidemiological investigation (which are by nature, observational not experimental), there is no such thing as a
perfect epidemiology study. Some epidemiologic investigations have been complicated by
such factors as failure to adequately control exposure to other known carcinogens or an
inadequate period of observation to allow for full expression of cancers with long latent
periods. Other studies suffer from inadequate design, resulting in problems such as small
study size or biased selection of case and control populations or poor assessment of estimated exposure. Exposure assessment that relies on data that are incomplete, inaccurate,
or a surrogate for the actual exposure of interest can lead to flawed conclusions. In retrospective studies in particular, the use of questionnaires that relied on peoples recollections
to estimate exposure can be a particularly problematic, especially if there is a high likelihood of recall bias among cases compared to controls. For example, the case subjects may
have more reliable memories than the control subjects because they have been searching
for a plausible explanation of the cause of their disease. These methodological issues notwithstanding, epidemiology has made invaluable contributions to public health, espeTABLE 20-9STATISTICAL POWER CALCULATIONS FOR A H
YPOTHETICAL
STUDY IN WHICH THE BASELINE CANCER RISK, FOR AN

UNSPECIFIED SUBSET OF CANCER SITES, IS KNOWN TO BE 10%,
AND THE UNKNOWN RADIATION-RELATED EXCESS RISK IS 10%
AT 1 Gy AND PROPORTIONAL TO DOSE BETWEEN 0 AND 1 Gy
RADIATION DOSE
EXCESS RISK
TOTAL RISK
POPULATION SIZE N
1 Gy
10%
20%
80
100 mGy
1%
11%
6,390
10 mGy
0.1%
10.1%
620,000
1 mGy
0.01%
10.01%
61.8 million
N, the population size needed for 80% power to detect the excess risk at the 5% significance level.
Source: Adapted from International Commission on Radiological Protection. Low-dose extrapolation of
radiation-related cancer risk. ICRP Publication 99. Ann ICRP 2006;35:1140.
802
cially in cases where the exposures, such as in smoking, resulted in widespread adverse
public health consequences. The situations where epidemiological investigations have
the most difficulty are where the risks are small compared to the normal incidence of the
disease and where exposures in cases and controls are difficult to quantify. Table 20-10
summarizes details of some of the principal epidemiologic investigations on which current
dose-response estimates are based. Several excellent overviews on radiation epidemiology
TABLE 20-10SUMMARY OF MAJOR EPIDEMIOLOGIC INVESTIGATIONS
THAT FORM THE BASIS OF CURRENT CANCER DOSERESPONSE ESTIMATES IN HUMAN POPULATIONS
POPULATION AND EXPOSURE
EFFECTS OBSERVED
STRENGTHS AND LIMITATIONS
A-bomb survivors: The LSS of

the Japanese A-bomb survivors
has provided detailed epidemiological data from a study
of this population for about
50 y. Three cohorts currently
being studied are: (1) A cancer
(and noncancer) incidence and
mortality study of ~105,000
residents of Hiroshima and
Nagasaki (1950) with doses
ranging from 0 (e.g., not in the
city at time of the bombing)
to 4 Gy (42% received a dose
between 5 and 100 mGy);
(2) those exposed in utero
(~3,600) and (3) F1 generation children of those exposed
(~77,000). The cancer incidence
and mortality assessment
through 1998 has been completed. Mean organ doses have
been calculated for 12 organs.
Risk estimates were revised in
2006 by the National Academy
of Sciences / National Research
Council Committee on the
Biological Effects of Ionizing
Radiation. Their reanalysis of
the scientific data on low-dose
radiation health effects was
undertaken in light of a reassessment of the doses received
by the Japanese atomic-bomb
survivors, referred to as the
DS02 dose estimate, (Young,
2005), additional information on nontargeted effects
of radiation (e.g., bystander
effect, low-dose hypersensitivity) as well as an additional
decade of follow-up of the
A-bomb survivors.
A-bomb survivor data demonstrates an undeniable increase

in cancer for doses .100 mSv
(some say 50 while others
say 200 mSv). Excess risks of
most cancer types have been
observed, the major exceptions
being chronic lymphocytic
leukemia, multiple myeloma,
non-Hodgkins lymphoma,
pancreatic, prostate and
gall bladder cancer. A total
of 853 excess solid cancers
from a total of 17,488 cases
are thought to have been
induced by radiation exposure.
Table20-11 lists, for each dose
category, the observed and
expected numbers of cancers,
the excess number of cancers,
and the percent of cancers
that can be attributed to
radiation exposure (attributable fraction). The small but
statistically significant increase
in the 5100 mGy exposure
group is of particular interest
in medical imaging as it is similar to organ dose experienced
in many diagnostic imaging
studies. In more than 50 y
of follow-up of the 105,427
atomic-bomb survivors, the
percent of cancers attributed
to their radiation exposure
is 10.7%. Estimates of the
site-specific solid cancer risks
are shown in Figure 20-25. The
influence of sex and age at
the time of exposure, and the
risk as a function of attained
age following exposure can be
very significant and examples
of their influence are shown in
Figure 20-26.
The analysis of the data from

the atom-bomb survivors
cohort is the single most
important factor that has
influenced current radiationinduced cancer risk estimates.
The population is large and
there is a wide range of doses
from which it is possible to
determine the dose-response
and the effects of modifying
factors such as age on the
induction of cancer. Data at
high doses are limited; thus
the analysis only included
individuals in whom the
doses were 2 Gy or less. The
survivors were not representative of a normal Japanese
population insofar as many
of the adult males were away
on military service while
those remaining presumably
had some physical condition
preventing them from active
service. In addition, the children and the elderly perished
shortly after the detonation
in greater numbers than did
young adults, suggesting the
possibility that the survivors
may represent a hardier
subset of the population.
Another important uncertainty is the transfer of sitespecific cancer risk estimates
to the U.S. population, based
on results obtained on the
LSS population, for cancers
with substantially different
baseline incidence rates.
803
TABLE 20-10SUMMARY OF MAJOR EPIDEMIOLOGIC INVESTIGATIONS
THAT FORM THE BASIS OF CURRENT CANCER DOSERESPONSE ESTIMATE IN HUMAN POPULATIONS (continued)
POPULATION AND EXPOSURE
EFFECTS OBSERVED
STRENGTHS AND LIMITATIONS
Ankylosing spondylitis: This

cohort consists of ~14,000
patients treated with radiotherapy to the spine for ankylosing spondylitis throughout the
United Kingdom between 1935
and 1954. Although individual
dose records were not available
for all patients, estimates were
made ranging from 1 to 25 Gy
to the bone marrow and other
various organs.
Mortality has been reported

through 1982, at which point
727 cancer deaths had been
reported. Excess leukemia
rates were reported from
which an absolute risk of 80
excess cases/Gy/y per million
was estimated.
This group represents one of

the largest bodies that has
provided data on radiationinduced leukemia in humans
for which fairly good dose
estimates exist. Control
groups were suboptimal,
however, and doses were
largely unfractionated. In
addition, only cancer mortality (not incidence) was available for this cohort.
Postpartum mastitis study: This

group consists of ~600 women,
mostly between the ages of
20 and 40 y, treated with
radiotherapy for postpartum
acute mastitis in New York in
the 1940s and 1950s for which
~1,200 nonexposed women
with mastitis and siblings of
both groups of women served
as controls. Breast tissue doses
ranged from 0.6 to 14 Gy.
Forty-five year follow-up

identified excess breast
cancer in this population as
compared with the general
female population of New
York.
A legitimate objection to
using the data from this
study to establish radiationinduced breast cancer risk
factors is the uncertainty as
to what effect the inflammatory changes associated with
postpartum mastitis and the
hormonal changes due to
pregnancy have on the risk of
breast cancer.
Radium dial painters: Young

women who ingested radium
(Ra-226 and Ra-228 with
half-lives of ~1,600 and 7 y,
respectively) while licking their
brushes (containing luminous
radium sulfate) to a sharp
point during the application
of luminous paint on dials and
clocks in the 1920s and 1930s.
Over 800 were followed.
Large increase in osteogenic

sarcoma. Osteogenic sarcoma
is a rare cancer (incidence,
~5 per 106 population).
RR in the population was
.100s. No increase was
seen below doses of 5 Gy but
a sharp increase was noticed
thereafter.
One of only a few studies

that analyzed the radiocarcinogenic effectiveness of
internal contamination with
high-LET radiation in humans.
Thorotrast: Several populations were studied in which

individuals were injected
intravascularly with an x-ray
contrast medium, Thorotrast,
used between 1931 and 1950.
Thorotrast contains 25% by
weight radioactive colloidal
Th-232 dioxide. Th-232 is an
alpha emitter with a half-life
of ~14 billion y.
Particles were deposited in the

reticuloendothelial systems.
Noted increase in number
of cancers, particularly liver
cancer (angiosarcoma, bile
duct carcinomas, and hepatic
cell carcinomas) and leukemia.
Evaluation of the data resulted
in estimates of alpha radiationinduced liver cancer risk of ~8
102 per Gy which appears to
be linear with dose. Alpha RBE
~20. An increase in leukemia
was also seen, however, the
RBE was much lower (~1).
Dose estimates are fairly

good. However, the extent to
which the chemical toxicity of
the Thorotrast may have influenced the risk is not known.
Thorotrast administration
resulted in chronic alpha particle irradiation from radionuclides in the thorium decay
series. Organs of deposition of
Th-232 and from the daughter products of radon-220 in
the lungs and of radium-224
and its decay products in the
skeletal system.
Source: Adapted and updated and from National Academy of Sciences/National Research Council Committee
on the Biological Effects of Ionizing Radiation. The health effects of exposure to low levels of ionizing radiation (BEIR V). Washington, DC: NAS/NRC, 1990.
804
TABLE 20-11 SOLID CANCERS CASES BY DOSE CATEGORY

DOSE CATEGORYa
SUBJECTS
OBSERVED
BACKGROUNDb
FITTED EXCESSb
ATTRIBUTABLE
FRACTION (%)
,0.005
60,792
9,597
9,537
0.0050.1
27,789
4,406
4,374
81
1.8
0.10.2
5,527
968
910
75
7.6
0.20.5
5,935
1,144
963
179
15.7
0.51
3,173
688
493
206
29.5
12
1,647
460
248
196
44.2
24b
564
185
71
111
61
Total
105,427
17,448
16,595
853
10.7
Weighted colon dose in Gy.

Note Estimates of background and fitted excess cases are based on an ERR model with a linear dose response
with effect modification by gender, age at exposure and attained age.
c
Note that the most reliable source of epidemiological data in the LSS include doses range up to 2Gy, beyond
which, errors in dose reconstruction and mortality from complications of the ARS provided limited radioepidemiological information.
Source: Modified from (Preston, 2007).
a
and perspective on the relative strength of the evidence supporting current risk estimates
for radiation-induced cancers and future challenges are available in the literature and are
highly recommended (Boice, 2006, 2011; UNSCEAR, 2008).
Estimates of Radiation-Induced Cancer Risks from

Low-Dose, Low-LET Radiation
Numerous national and international scientific organizations periodically report on
the state of scientific knowledge regarding the carcinogenic risk and other biological
All solid cancers
Esophagus
Stomach
Colon
Rectum
Liver
Gall bladder
Pancreas
Lung
Breast
Uterus
Ovary
Prostate
Bladder
Other solid
Deaths
9335
291
2867
478
370
1236
328
407
1264
275
518
136
104
150
911
P value
P<0.001
P<0.001
P<0.001
P<0.001
P=0.14
P<0.001
P=0.007
P=0.29
P<0.001
P<0.001
P=0.20
P=0.004
P=0.42
P=0.02
P<0.001
0.2
0.0
0.5
1.0
1.5
ERR (per Sv)

FIGURE 20-25 Estimates of the site-specific solid cancer mortality ERR with 90% confidence intervals.
Except for gender-specific cancers (breast, ovary, uterus, and prostate), the estimates are averaged over gender. The dotted vertical line at 0 corresponds to no excess risk, while the solid vertical line indicates the genderaveraged risk for all solid cancers. Some cancer sites such as brain, testes, cervix, oral cavity, and kidney are not
included. (Source Preston DL, Shimizu Y, Pierce DA, et al. Studies of mortality of atomic bomb survivors. Report
13: Solid cancer and noncancer disease mortality: 19501997. Radiat Res 2003;160:381407).
805

2
Gender
1.5
1.5
Age at exposure
10
Female
Male
40
1
.5
.5
er
ro
id
th
O
Th
y
lid
So
om
ac
St
st
de
Bl
ad
Br
ea
ng
Lu
Li
ve
C
ol
on
de
r
om
ac
h
So
lid
Th
yr
oi
d
Li
ve
r
O
th
er
St
Bl
ad
ng
st
Lu
Br
ea
r
C
ol
on
Attained age
75
50
1.5
1
.5
on
ol
C
st
ve
Li
ng
ea
Br
Lu
lid
ac
So
d
oi
yr
om
St
th
er
Th
Bl
ad
de
FIGURE 20-26 Comparison of gender (A), age-at-exposure (B), and attained-age (C) effects on standardized ERR1 Gy for selected sites and all solid cancers in the LSS. The ERR estimates for the other category are based on the results of analyses of the 5,396 cancer cases not included in the sites explicitly
considered here. The gender-specific estimates correspond to the fitted ERR per Gy at age 70 for a person exposed at age 30. The age-at-exposure specific estimates are gender-averaged ERR estimates at age
70 after exposure at age 10 (red bar) or age 40 (green bar). Attained-age-specific estimates are gender averaged ERR estimates at ages 50 (red bar) and 75 (green bar) after exposure at age 30. Within each panel, the
sites are ordered based on the magnitude of the ratio of the effect pairs (Preston, 2007).
effects of ionizing radiation. Most recently, these include the United Nations Scientific
Committee on the Effects of Atomic Radiation (UNSCEAR, 2006, 2008, 2009, 2011)
report; the National Research Council, Committee on the Biological Effects of Ionizing
Radiations (BEIR VII) report entitled Health Effects of Exposure to Low Levels of Ionizing
Radiation in 2006 (BEIR, 2006); the U.S. Environmental Protection Agencys report on
Radiogenic Cancer Risk Models and Projections for the U.S. Population (EPA, 2011); as well
as reports focused on specific radiation health and radiation protection related topics
by the NCRP, ICRP, and others. While there are some differences in the interpretation of
specific aspects in the scientific literature among these various expert bodies, it is fair to
say that there is general agreement on the most important aspects of radiation-induced
risks, including the magnitude of the risk for the general population and its uncertainty
at low doses. General points of agreement that are consistent with the results from the
A-bomb survivor LSS include (1) risk varies according to cancer site; (2) risk is greater
when exposures occur at younger ages; (3) risk is greater for females than males; (4) solid
cancer risk is consistent with a linear function of dose when all cancers are combined;
(5) leukemia risk is consistent with a nonlinear function of dose; (6) with the exception
of leukemia for which there is a fairly well-defined risk interval following exposure, risk
remains elevated for 50+ years after exposure; and (7) there is no convincing epidemiological evidence of radiation causing genetic (inherited) effects. In as much as there is
not a major divergence of opinions expressed in the reports cited above, information
on radiation-induced cancer risks and genetic effects will be presented based on the
analysis and perspectives contained in the BEIR VII committee report.
806
BEIR VII Report

The BEIR VII report was an update of their previous 1990 report (BEIR V) on the same
topic and a report focused on risk from radon exposure (BEIR VI), which was published
in 1999. The primary objective of the report was to provide a comprehensive reassessment of the health risks resulting from exposures to low-dose, low-LET radiation in
humans. The report provided updated risk estimates based on a review of the relevant
scientific evidence from epidemiology and the vast array of data from laboratory experiments that included everything from long term animal studies to molecular mechanisms
of radiation-induced damage and response. For the purpose of the report, low dose
was defined as less than 100 mGy. The report also considered chronic exposure, defined
as dose rates less than 0.1 mGy/min, irrespective to total dose. The report provided risk
estimates of cancer incidence and mortality from radiation in the U.S. population as well
as population and organ-specific risk estimates adjusted for gender, age at exposure, and
time interval following exposure including lifetime risks. Heritable effects and risk from
in utero radiation exposure were also considered. For most cancer risk estimates, the
BEIR VII committee relied on the most recent cancer incidence and mortality data from
the A-bomb survivor LSS. For breast and thyroid cancer, the committee used a pooled
analysis of data from the LSS and studies of medically exposed persons.
Modeling of Epidemiological Data

Within the context of the limitations of epidemiology described earlier, scientists
have developed dose-response models to predict the risk of cancer in human populations from exposure to low levels of ionizing radiation. While the data from the LSS
are without doubt the most robust data set from which cancer risk estimates can be
made, their use requires answers to several important questions such as (1) how can
the effects demonstrated at high dose and high dose rate be compared to low dose
and low dose rate exposures typical of most medical, occupational, and environmental exposures; (2) what is the effect of age at time of exposure on risk estimates;
(3) how well do the risk estimates based on exposure in the Japanese population
translate to what might be expected in the US population; (4) how representative
is a population exposed in 1945 in a war-torn country with nutritional and other
deficiencies to a modern healthy population of today?
Dose-Response Models
Several models have been proposed to characterize dose-response relationships of
radiation exposure in humans. The shapes of the dose-response curves have been characterized as linear nonthreshold (LNT), linear-quadratic (LQ), and threshold (Fig.20-27).
The two LNT extrapolations represent the high dose and dose rate (black) and low
dose (orange) components of the dose-response data. The LQ dose-response curve
(red) demonstrates a reduced effectiveness for radiogenic cancer induction at lower
dose and higher effectiveness at higher dose that eventually flattens out, reflecting
doses associated with substantial cell killing.
In addition to the experimental evidence previously mentioned, there is also epidemiological evidence for a threshold for some cancers (e.g., blue dashed line in
Fig. 20-27). For example, a threshold model provides the best fit for the osteogenic
sarcoma risk among dial painters who had substantial internal radium contamination (discussed previously in this chapter). Other cancers, for which the evidence
for radiation-induced risk is consistent with a threshold, include bone, soft tissue
sarcoma, rectum, and nonmelanoma skin cancer. In addition, there are sites for
Cancer Incidence
807
Linear No-Threshold (high dose rate)

Linear Quadratic
Linear No-Threshold (low dose rate)
Linear Model with a Threshold
Radiation Dose
FIGURE 20-27 Possible dose response relationship for radiation-induced cancer. (Adapted from BEIR. Health
risks from exposure to low levels of ionizing radiation: BEIR VII, Phase 2. Committee to Assess Health Risks from
Exposure to Low Levels of Ionizing Radiation, Board of Radiation Effects, Research Division on Earth and Life
Studies, National Research Council of the National Academies. National Academy of Sciences, Washington,
DC: National Academies Press, 2006.)
which there is no convincing evidence for a radiation-induced increase in cancer risk

(e.g., prostate, pancreas, testes, cervix).
However, even though there is evidence for a threshold dose for some types of
radiation exposure in specific tissues below which no radiogenic cancer risk is evident, the evidence supporting this model is not sufficient to be accepted for general
use in assessing cancer risk from radiation exposure. Both the ICRP and BEIR VII
committee concluded that, although there are alternatives to linearity, there is no
strong evidence supporting the choice of another form of the dose response relationship. The statement in ICRP publication 99 captured the essence of the consensus
view in this regard: . while existence of a low-dose threshold does not seem to be
unlikely for radiation-related cancers of certain tissues, the evidence does not favor
the existence of a universal threshold. The LNT hypothesis, combined with an uncertain DDREF for extrapolation from high doses, remains a prudent basis for radiation
protection at low doses and low dose rates (ICRP, 2006).
Figure 20-28, adapted from BEIR VII, shows the point estimates (orange dots) of
ERRs of solid cancer incidence, (averaged over gender and standardized to represent
individuals exposed at age 30 who have attained age 60) for specific dose intervals
from the A-bomb survivor LSS. Vertical lines represent the 95% confidence intervals
around each point estimate. The orange solid line and the green dashed line are linear
and LQ dose-response models (respectively) for ERR of solid tumors, estimated from
all subjects with doses in the range 0 to 1.5 Sv. The LQ dose-response function was discussed previously in the context of cell survival curves. Here the probability of an effect
(cancer induction in this case) is proportional to the sum of two coefficients, one that
is proportional to dose (aD) and one that is proportional to square of the dose (bD2).
Just as with the cell survival curves, the linear and quadratic components represent the
response to low dose and high dose and dose rate effects, respectively. The degree of
curvature is the ratio of the quadratic and linear coefficients. These coefficients can be
estimated for different cancers reflecting their unique dose-response relationships.
As discussed previously, the biological effectiveness of the radiation-induced
damage to cells and tissues generally decreases at lower dose and lower dose rates.
This effect is due, at least in part, to the ability to repair damage during low dose
Linear Fit 01.5 Sv Solid Tumor with DDREF 1.5

Linear Fit 01.5 Sv Solid Tumor
Linear-Quadratic Fit 01.5 Sv Solid Tumor
Linear-Quadratic Fit Leukemia
0.5
0.0
Linear-Quadratic Fit Leukemia
5
4
3
2
Law Dose Range
1.0
Law Dose Range
Excess Relative Risk of

Solid Tumors and Leukemia
1.5
Excess Relative Risk of Leukemia
808
1
0
0.0
0.5
1.0
1.5
Radiation Dose (Sv)
2.0
0.0
0.5
1.0
1.5
Radiation Dose (Sv)
2.0
FIGURE 20-28 Dose-response models of ERR for solid cancer and leukemia from cancer incidence data of
Japanese A-bomb survivors. (Adapted from BEIR VII, 2006).
exposure or between exposures in the case of a higher total dose delivered in smaller
dose fractions. A DDREF has been used to convert risk from exposures at high dose
and high dose rate to estimates of risk at low doses for the purposes of radiation
protection. The BEIR VII committee applied a DDREF of 1.5 to adjust the response
observed in the high dose range making it roughly equivalent to the line representing
the LNT low-dose response (black dashed line in Fig. 20-28). This line represents an
extension of the linear portion (i.e., zero-dose tangent) of the LQ model (seen more
clearly as the low-dose LNT line in Fig. 20-27). The value of the DDREF was derived
based on a 95% confidence interval of a Bayesian statistical analysis of the solid cancer incidence data from the A-bomb survivor LSS, as well as the results from selected
laboratory animal data. The BEIR VII committee, however, noted the uncertainty in
their estimate of a DDREF because of the substantial inconsistency and imprecision
in the animal data and because the DDREF estimate was particularly sensitive to the
selection of the dose range used for estimation, the particular studies chosen for analysis, and the approach used for estimating curvature that is presumed to be the same
across all studies. While it would be equally correct to use the LQ fit to the solid cancer risk data, the BEIR VII committee chose to use the DDREF-adjusted linear doseresponse model for cancer risk estimates with the exception of leukemia, for which
the LQ model was a better fit. In the absence of a superior dose-response model, the
decision was made because it was simple to apply and is widely used by other organizations for this purpose, albeit with a somewhat different DDREF value (e.g., DDREF
of 2 was used by UNSCEAR and ICRP). Note that in the low-dose range of interest
in medical imaging and occupational exposure to radiation (less than 100 mSv), the
difference between the linear fit to the solid tumor data (incorporating a DDREF
1.5black dashed line) and the LQ fit to the same data (green dashed line) is relatively
small compared to the 95% confidence intervals. The red lines in the left and right
panels of Figure 20-28 show the ERR for leukemia which best fit a LQ dose-response
model. The left panel shows the ERR for leukemia plotted on the same ERR scale as
the solid cancer risk data up to approximately 0.7 Sv while the panel on the right
shows the same dose-response up to 2 Sv on a broader ERR scale. One can appreciate
the greater ERR per unit dose (i.e., the degree of curvature) observed for this cancer.
Multiplicative and Additive Risk Models: (Transport of Risk

Between Populations)
Previous estimates of radiation-induced cancer have employed both the multiplicative and
the additive risk models. The multiplicative risk-projection model (also referred to as the
809
relative risk model) is based on the assumption that the excess cancer risk increases in
proportion to the baseline cancer rate. Thus, the multiplicative risk model predicts that,
after the latent period (e.g., 10 years), the excess risk is a multiple of the natural age-specific risk for the specific cancer and population in question (Fig. 20-29A). The alternative
additive (or absolute) risk model (expressed in in terms of EAR) is based on the assumption that, following the latent period, the excess cancer rate is constant and independent
of the spontaneous population and age-specific natural cancer risk (see Fig. 20-29B).
While neither of these two simple models appears adequate to completely describe the
risk of radiation-induced cancer, the multiplicative risk model is consistent with the scientific evidence that radiation acts predominantly as an initiator, rather than a promoter,
of carcinogenesis. In addition, the copious epidemiological evidence, which indicates that
exposure of children to radiation carries a greater risk than exposures later in life, further
supports its role as an initiator rather than promoter. The multiplicative risk model was
used by the BEIR VII committee for deriving tissue-specific solid cancer risk estimates as
a function of gender, age at time of exposure, and time elapsed since exposure.
The multiplicative and additive risk-projection models can be used to transport
risk calculated from one population to another dissimilar population. This is a critical
issue, insofar as the LSS of the Japanese A-bomb survivors serves as the foundation
of many radiation risk estimates and radiation protection regulations. The problem
is especially important for U.S. population based radiation risk estimates because the
natural incidence of several types of cancer in the U.S. and in Japan are very dissimilar.
For example, the incidence of breast cancer in the U.S. is 3 times that of Japan while
the incidences of liver and stomach cancer are 7.5 and 10 times higher respectively in
A
Multiplicative Risk Model

Incidence in exposed population
Incidence
Spontaneous incidence (unexposed)
Radiation
exposure
Latency
X+
90
Incidence
Additive Risk Model
Radiation
exposure
Latency
X+
Age
90
FIGURE 20-29 Comparison of multiplicative and additive risk models. Radiation-induced risk increments are seen
after a minimal latent period ( ); X is
age at exposure. A. In the multiplicative
risk model, the excess risk is a multiple
of the natural age-specific cancer risk
for a given population which increases
with age. B. The additive risk model
is shown in which a fixed incremental
increased risk is added to the spontaneous disease incidence which is assumed
to be constant over the remaining
lifetime. The multiplicative risk model
predicts the greatest increment in incidence at older ages.
810
TABLE 20-12BASELINE CANCER INCIDENCE RATES IN
UNITED STATES AND JAPAN (FEMALES)a

U.S.
Japan
All
280
185
Stomach
3.5
34
Colon
22
17
Liver
1.3
9.8
Lung
34
12
Breast
89
30
Bladder
5.9
2.6
Source: Correa CN, Wu X-C, Andrews P, et al., eds. Cancer incidence in five continents, vol VIII. Lyon, France: IARC Scientific Publications No. 155, 2002.
a
incidence rates per 100,000 women.
Japan than in the U.S. (Table 20-12). The previous BEIR committee (BEIR V) based its
estimates on multiplicative risk transport, where it is assumed that the excess risk due
to radiation is proportional to baseline cancer risks. BEIR VII took a hybrid approach
using the multiplicative risk model for some cancers (e.g., thyroid); the absolute risk
model for others (e.g., breast); and, for most, a weighted average between the values
calculated using the absolute and multiplicative risk models.
Age and Gender Specific Radiation Risk Estimates

A summary of the BEIR VII preferred estimates of lifetime attributable risk of solid
cancer (incidence and mortality) along with their 95% confidence intervals for different exposure scenarios is shown in Table 20-13. Tables of lifetime attributable
TABLE 20-13BEIR VII PREFERRED ESTIMATES OF LIFETIME ATTRIBUTABLE
RISK OF SOLID CANCER INCIDENCE AND MORTALITY WITH

95% CONFIDENCE INTERVALSa
EXPOSURE
SCENARIO
0.1 Gy to
population of
mixed ages
INCIDENCE
MEN
800 (4001,590)
MORTALITY
WOMEN
1,310 (6902,490)
MEN
410 (200830)
WOMEN
610 (3001,230)
0.1 Gy at age 10 1,330 (6602,660)
2,530 (1,2904,930) 640 (3001,390)
1,050 (4702,330)
0.1 Gy at age 30 600 (2901,290)
1,000 (5002,020)
490 (250950)
320 (150650)
0.1 Gy at age 50 510 (2401,100)
680 (3501,320)
290 (140600)
420 (210810)
1 mGy/y
550 (2801,100)
throughout life
970 (5101,840)
290 (140580)
460 (230920)
10 mGy/y from
ages 18 to 65
2,600 (1,2505,410) 4,030 (2,0707,840) 1,410 (7002,860) 2,170 (1,1304,200)
Cancer incidence and mortality per 100,000 exposed persons. 95% Confidence intervals shown in parentheses.
Source: BEIR. Health risks from exposure to low levels of ionizing radiation: BEIR VII, Phase 2. Committee to
Assess Health Risks from Exposure to low levels of Ionizing Radiation, Board of Radiation Effects, Research
Division on Earth and Life Studies, National Resarch Council of the National Academies. National Academy of
Sciences, Washington, DC: National Academies Press, 2006.
811
cancer incidence and mortality risk by age at exposure and cancer site are provided
in the BEIR VII report. The cancer incidence data from that document are shown
below in Table 20-14 and the mortality data table can be found in Chapter 11,
Table 11.6.
The risks in these tables are expressed as the number of additional cases per
100,000 per 100 mGy to the specified tissue. From this information, organ-specific
risk estimates can be derived. For example, if a 10-year-old girl received a 20 mGy
TABLE 20-14LIFETIME ATTRIBUTABLE RISK OF SITE-SPECIFIC CANCER
INCIDENCE
NUMBER OF CASES PER 100,000 PERSONS EXPOSED TO A SINGLE DOSE OF 0.1 Gy
AGE AT EXPOSURE (y)
Cancer Site
10
15
20
30
40
50
60
70
80
Men
Stomach
76
65
55
46
40
28
27
25
20
14
Colon
336
285
241
204
173
125
122
113
94
65
30
Liver
61
50
43
36
30
22
21
19
14
Lung
314
261
216
180
149
105
104
101
89
65
34
93
80
67
57
48
35
35
33
26
14
5
23
Prostate
Bladder
209
177
150
127
108
79
79
76
66
47
1,123
672
503
394
312
198
172
140
98
57
23
Thyroid
115
76
50
33
21
0.3
0.1
0.0
All solid
2,326
1,667
1,325
1,076
881
602
564
507
407
270
126
237
149
120
105
96
84
84
84
82
73
48
2,563
1,816
1,445
1,182
977
686
648
591
489
343
174
Other
Leukemia
All cancers
Women
Stomach
101
85
72
61
52
36
35
32
27
19
11
Colon
220
187
158
134
114
82
79
73
62
45
23
Liver
28
23
20
16
14
10
10
Lung
733
608
504
417
346
242
240
230
201
147
77
Breast
1,171
914
712
553
429
253
141
70
31
12
Uterus
50
42
36
30
26
18
16
13
Ovary
104
87
73
60
50
34
31
25
18
11
Bladder
212
180
152
129
109
79
78
74
64
47
24
Other
1,339
719
523
409
323
207
181
148
109
68
30
Thyroid
634
419
275
178
113
41
14
0.3
0.0
All solid
4,592
3,265
2,525
1,988
1,575
1,002
824
678
529
358
177
185
112
86
76
71
63
62
62
57
51
37
4,777
3,377
2,611
2,064
1,646
1,065
886
740
586
409
214
Leukemia
All cancers
Source: Adapted from BEIR VII Table 12 D-1, BEIR. Health risks from exposure to low levels of ionizing
radiation: BEIR VII, Phase 2. Committee to Assess Health Risks from Exposure to Low Levels of Ionizing
Radiation, Board of Radiation Effects, Research Division on Earth and Life Studies, National Research
Council of the National Academies. National Academy of Sciences, Washington, DC: National Academies
Press, 2006.
812
breast dose from a chest CT exam, the lifetime risk of being diagnosed with breast
cancer as a result of the exam would be:
From Table 20-14, the radiation-induced breast cancer risk at age 10 is 712 cases per
100,000 per 100 mGy. Thus the risk would be calculated as (712) (20 mGy/100 mGy)
142 per 100,000 or approximately 1 in 700.
If the same dose were received by a 50-year-old woman the risk would drop by
approximately a factor of 10 (~1 in 7,000). It is important to understand that the
risk coefficients presented in these tables have large uncertainties associated with
them and there are many other risk factors for breast cancer (discussed below). Furthermore, the evidence that radiation causes breast cancer at 100 mSv is equivocal
and this uncertainty is even greater for doses less than 20 mSv, implying that the
statistical uncertainty of risk estimates at these low doses is large and the possibility
of no increase in risk cannot be excluded. Nevertheless they can be used to provide
a perspective of the magnitude of the risk engendered from the dose received for a
particular exam in a given patient population. A number of such risks estimates have
been made in recent years for a variety of imaging studies. The result from one such
study evaluating the change in dose and risk with age of exposure for CT exams of
the head and abdomen is shown in Figure 20-30. As discussed in Chapter 21, the
use of medical imaging using ionizing radiation has increased dramatically in the last
20 years. There has been a growing awareness of the potential for overutilization of
this technology and the attendant radiation risk to patients (Brenner, 2007; Frazel,
2009). This issue is particularly troubling for the pediatric patient population who
have inherently higher cancer risks from a given dose and (depending on the imaging
study) may receive higher organ doses than adults for the same procedure (Brenner,
2001; Brody, 2007). Efforts to educate physicians and patients about these issues
continue and will no doubt be necessary for many years to come. Optimization of
imaging procedures in this context includes not only the technical factors related to
image acquisition but also (and often most importantly) the appropriateness of the
requested examination to resolve the particular clinical question. The pressure in
some cases from parents who may insist on a CT exam for the child being evaluated
for relatively minor head injury adds to this problem. The use of clear evidence based
guidelines and appropriateness criteria for medical imaging procedures is an important quality care goal and their implementation in an emergency department setting
when evaluating pediatric patients is especially important (Kuppermann, 2009).
Estimates of the total cancer incidence and mortality risk as a function of sex and
age at exposure are presented graphically in Figure 20-31. The increased cancer risk
associated with exposure to radiation of children and infants is easily appreciated. Compared to the risk to adults, the cancer risk for a 12-month-old infant is three to four
times higher. The increased risk of radiation exposure in females compared to males at
all ages is also shown; however, the magnitude of the difference decreases with age.
Population Radiation Risk Estimates

The population averaged radiation-induced cancer incidence and mortality risk estimates, calculated from data provided in the BEIR VII report, are shown in Table 20-15.
The risk estimates of the International Commission on Radiological Protection (ICRP),
(derived from evaluation of the radiation-induced cancer incidence and mortality
data), are in general agreement with those in the BEIR VII report. As mentioned earlier
the ICRP risk estimates were made for the purpose of radiation protection (see Chapter
3) and are not directly comparable to the BEIR VII cancer risk estimates. The current
813
Abdominal CT, 240 mAs

Estimated Organ Dose (mGy)
Estimated Organ Dose (mGy)
Head CT, 340 mAs

100
Brain
Bone
Bone marrow
Thyroid
80
60
40
20
0
10
20
30
40
50
60
70
Age at Time of CT Study (yr)
Total
Brain
Leukemia
0.06
0.04
0.02
10
15
10
5
0
10
20
30
40
50
60
70
Abdominal CT, 240 mAs
0.08
20
Estimated Lifetime Attributable

Risk of Death from Cancer (%)
Estimated Lifetime Attributable

Risk of Death from Cancer (%)
25
Head CT, 340 mAs
0.00
Stomach
Liver
Ovaries
Colon
Bone marrow
30
20
30
40
50
60
70
Total
Digestive
Other
Leukemia
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
10
20
30
40
50
60
70
FIGURE 20-30 Estimated organ doses and lifetime cancer risk with age of exposure for CT exams of the
head and abdomen. A, B show estimated radiation doses for selected organs from a single CT scan of the
head or the abdomen. As discussed in Chapter 10, for a given mAs, pediatric doses are typically larger than
adult doses, because there is less self shielding of organs during the tube rotation. Exposure parameters and
other aspects of imaging studies should be optimized for each patient. However, it is especially important for
children due to their increased risk of radiogenic cancer and often higher organ doses. Optimization of imaging procedures in this context includes not only the technical factors related to the image acquisition but also
(and often most importantly) the appropriateness of the requested examination to resolve the particular clinical question. Panels C, D show the corresponding estimated lifetime percent cancer mortality risk attributable
to the radiation from a single CT scan; the risks (both for selected individual organs and overall) have been
averaged for male and female patients. One can appreciate that even though doses are higher for the head
CT scans, the risks are higher for abdominal scans because organs exposed are more sensitive than the brain
to radiation-induced cancer. It should be noted that the mAs (and resultant doses) used in these examples are
from the source cited below and are higher than typical doses for similar exams performed with image acquisition parameters optimized for smaller patients. Contrary to current practice, these data assume that no tube
current adjustment for patient size has been applied, which of course would substantially reduce the dose (and
thus risk) in the pediatric population. (From Brenner DJ, Hall EJ. Computed tomographyan increasing source
of radiation exposure. N Engl J Med 2007;357:22772284.)
ICRP risk estimates of radiation-induced detriment1 (e.g., cancer and genetic effects)
at low dose are 4.2% per Sv for a population of adult workers and 5.7% per Sv for
the whole population (which includes more radiosensitive subpopulations, such as
children). The majority of this detriment is assigned to a cancer risk of 4.1% and 5.5%
per Sv for adult workers and the general population, respectively. Even though the
definitions and methodology of calculating radiation-induced cancer risk are different
for the ICRP and the BEIR VII committee, the ICRP population estimate of cancer detriment from radiation exposure (5.5%/Sv) corresponds fairly well to BEIR VII lifetime
attributable cancer mortality projections averaged for the U.S. population (5.7%/Sv).
1
The total harm to health experienced by an exposed group and its decendants as a result of the groups
exposure to a radiation source. Detriment is a multimentional concept. Its principal components are the
stochastic quantities: probability of attributable fatal cancer, weighted probability of attributable non-fatal
cancer, weighted probability of severe heritable effects, and length of life lost if the harm occurs.
814

5.0
Male Cancer Incidence
4.5
Male Cancer Mortality
Excess Cancers
3.5
3.0
2.5
AGE
F/M
Incidence
Risk
~4.3
1.9
~3.3
30
1.4
3.2
50
1.3
Risk at 1
vs 30 yrs
Female Cancer Mortality

4.0
Effect of Sex
At Age of Exposure
Effect of Age
At Exposure M&F
Female Cancer Incidence
Female Cancer
Incidence
Male Cancer
Incidence
Female Cancer
Mortality
Male Cancer
Mortality
2.8
2.0
1.5
11.4%/Sv
1.0
5.7%/Sv
0.5
0.0
10
20
30
40
50
60
70
80
Age at Exposure
FIGURE 20-31 Lifetime radiation cancer incidence and mortality as a function of sex and age at exposure.
Excess cancer cases per 1,000 following a whole body dose 10 mSv (multiply by 10 to convert to percent per
Sv). The table inserts shows the relative effect on cancer incidence and mortality for exposures at ages 1 and 30
years old (left) and the relative increased radiogenic cancer risk for females compared to males as a function
of exposure at 1,30, and 50 years old (right). The population averaged cancer incidence of 11.4%/Sv (green
line) and mortality 5.7%/Sv (red line) as calculated from the BEIR VII data is superimposed over the age and
sex-specific risk estimates. (Adapted from BEIR. Health risks from exposure to low levels of ionizing radiation:
BEIR VII, Phase 2. Committee to Assess Health Risks from Exposure to Low Levels of Ionizing Radiation, Board
of Radiation Effects, Research Division on Earth and Life Studies, National Research Council of the National
Academies. National Academy of Sciences, Washington, DC: National Academies Press, 2006.)
To estimate risk to the general population from radiation exposure at low exposure
levels the approximate values of 11% and 6% per Sv, respectively for cancer incidence
and mortality can be used. As mentioned earlier, the lifetime probability of developing or dying from cancer in the United States (averaged for both sexes) is approximately 41% and 22%, respectively. According to the linear risk-projection model,
an acute exposure of 100 people to 100 mSv would add approximately 1additional
TABLE 20-15THE US POPULATION AVERAGED
RADIATION-INDUCED CANCER
INCIDENCE AND MORTALITY RISK
ESTIMATES IN PERCENT PER Sv
INCIDENCE
Female
Male
U.S. population average
MORTALITY
13.7
6.6
9.0
4.8
11.4
5.7
Source: Calculated from: Health risks from exposure to low levels of ionizing
radiation: BEIR VII, Phase 2. Committee to Assess Health Risks from Exposure
to Low Levels of Ionizing Radiation, Board of Radiation Effects, Research
Division on Earth and Life Studies, National Research Council of the National
Academies. National Academy of Sciences, Washington, DC: National Academies Press, 2006.
815
0 mSv
5-80 years
After
Exposure
100 people
41 Cancers
100 mSv
5-80 years
After
Exposure
100 people
42 Cancers
FIGURE 20-32 Cancer incidence for two identical groups of 100 people with age and sex distributions similar to the general population. An acute exposure to 100 mSv for each person in one group would be expected
to add approximately 1 additional cancer case (occurring anytime after a latent period of approximately 5 years)
to the 41 normally expected to occur in the identical unexposed group over the groups lifetime.
cancer case to the 41 normally expected to occur over the lifetime of a group with an
age and sex distribution similar to the general population (Fig. 20-32).
Of course there would be no way of identifying which person had the additional
cancer (if it occurred at all) that was caused by the radiation exposure and given the
small number of individuals exposed, the natural variation associated with the average
incidence of cancer would be larger than one. The number of exposed individuals and
the dose in the example above were chosen to provide a simple example of the LNT
dose-response relationship. However, it would be very unusual for such a group to
receive such a large whole body dose at one time. Using a more realistic dose from an
(albeit unlikely) exposure scenario, one could imagine an accidental release of radioactive material into the environment exposing a population of 10,000 to 10 mSv in 1 year
(i.e., ten times the annual public exposure limit in the U.S.). The LNT model would
project 11 additional cancer cases (approximately one half of which would be fatal) over
the life time of the population. This represents an increase of less than 0.3% above the
spontaneous cancer incidence rate. UNSCEAR and ICRP, however, argue against using
collective dose to predict future cancer deaths when the average dose to the population
is small and below the doses where excess risks have been convincingly demonstrated.
An additional perspective on cancer risks is presented in Table 20-16, which compares
adult cancers with respect to their spontaneous and radiation-induced incidence.
Cancer Risk for Selected Organs and Tissues

Leukemia
Leukemia is a relatively rare disease, with an incidence in the general U.S. population of approximately 1 in 10,000 per year. However, genetic predisposition to
816
TABLE 20-16SPONTANEOUS INCIDENCE AND SENSITIVITY OF VARIOUS
TISSUES TO RADIATION-INDUCED CANCER

SITE OR TYPE OF CANCER
SPONTANEOUS INCIDENCE
RADIATION SENSITIVITY
Most frequent radiation-induced cancers

Female breast
Very high
High in children and young women
Thyroid
Low
Very high in children, especially in

females, but very low in adults
Lung (bronchus)
Very high
Moderate
Leukemia
Moderate
Very high
Alimentary tract
High
Moderate
Less frequent radiation-induced cancers

Pharynx
Low
Moderate
Liver and biliary tract
Low
Moderate
Lymphomas
Moderate
Moderate
Kidney and bladder
Moderate
Low
Brain and nervous

system
Low
Low
Salivary glands
Very low
Low
Bone
Very low
Low
Skin
High
Low
Magnitude of radiation risk uncertain

Larynx
Moderate
Low
Nasal sinuses
Very low
Low
Parathyroid
Very low
Low
Ovary
Moderate
Low
Connective tissue
Very low
Low
Radiation risk not demonstrated

Prostate
Very high
Absent?
Uterus and cervix
Very high
Absent?
Testis
Low
Absent?
Mesothelium
Very low
Absent?
Chronic lymphocytic
leukemia
Low
Absent?
Source: Modified from Committee on Radiological Units, Standards and Protection: Medical Radiation:
A Guide to Good Practice. Chicago, IL: American College of Radiology, 1985.
leukemia can dramatically increase the risk. For example, an identical twin of a leukemic child has a one in three chance of developing leukemia. Although it is rare in
the general population, leukemia is one of the most frequently observed radiationinduced cancers. Leukemia may be acute or chronic, and it may take a lymphocytic
or myeloid form. With the exception of chronic lymphocytic leukemia and viralinduced leukemia, increases in all forms of leukemia have been detected in human
populations exposed to radiation and in animals experimentally irradiated.
Within a few years after the detonation of the A-bombs, an increase in the incidence of leukemia was apparent in the survivor population. This evidence, together
with subsequent studies of medically and occupationally exposed cohorts, indicate
817
Incidence of Leukemia
Age < 15
1529
3044
45+
10
15
20
Time After Exposure (years)
25
FIGURE 20-33 Effect of age at the time of exposure on the incidence of leukemia (all forms except chronic
lymphocytic leukemia) among the A-bomb survivors. (From Ichuimaru et al. Incidence of leukemia in atomicbomb survivors, Hiroshima and Nagasaki 1950 to 1971, by radiation dose, years after exposure, age, and type
of leukemia. Technical Report RERF 10-76. Hiroshima, Japan: Radiation Effects Research Foundation, 1976.)
that excess leukemia cases can be detected as early as 1 to 2 years after exposure
and reach a peak approximately 12 years after the exposure. The incidence of leukemia is influenced by age at the time of exposure. The younger the person is at the
time of exposure, the shorter are the latency period and the period of expression
(Fig. 20-33).
Although the incidence of radiation-induced leukemia decreases with age at the
time of exposure, the interval of increased risk is protracted. The BEIR VII committees preferred model for leukemia risk was a LQ function of dose and, like models for
solid cancers, leukemia risk is expressed as a function of age at exposure. However,
the unique temporal wave-like character of the elevated risk for leukemia makes it
more useful to express risk as a function of time since exposure rather than attained
age (Fig. 20-34).
20
Excess relative risk per Sv
Age at exposure 10
Age at exposure 20
Age at exposure 30+
15
10
10
20
30
40
50
60
Years since exposure

FIGURE 20-34 Age-time patterns in radiation-associated risks for leukemia mortality. Curves are
sex-averaged estimates of the risk at 1 Sv for people exposed at age 10 (red line), age 20 (orange line), and
age 30 or more (blue line). (Adapted from BEIR VII.)
818
Another unique feature of the risk model is that there was no need to apply a
DDREF as the estimates are already based on a LQ function of dose. The BEIR VII
preferred model for leukemia estimates an excess lifetime risk from an exposure
to 0.1 Gy for 100,000 persons with an age distribution similar to that of the U.S.
population to be approximately 70 and 100, for females and males, respectively
(i.e., ~1% per Sv).
Thyroid Cancer
Distinct from the previously discussed cancer risk estimates, the preferred BEIR VII
models for thyroid and breast cancers are based on pooled analyses of the LSS data and
medically irradiated cohorts. Spontaneous and radiation-induced thyroid cancer shows
a greater difference by gender than most cancers, with women having a risk approximately two to three times as great as that for men (presumably because of hormonal
influences on thyroid function). Persons of Jewish and North African ancestry also
appear to be at greater risk than the general population. Thyroid cancer accounts for
only approximately 2% of the yearly total cancer incidence. In addition, the mortality is
low, resulting in only approximately 0.2% of all cancer deaths each year. Even though
there is a large relative risk for thyroid cancer from radiation exposure in childhood, the
absolute risks (even in children and young adults) are lower than for breast, colon and
lung cancer. There is also a rapid decrease in radiogenic cancer risk with age, with the
lifetime absolute risk per Gy for thyroid cancer at age 40 and beyond lower than the
risk to all but a few other organs and tissues. Most studies do not indicate any elevation
in risk when exposure occurs in adulthood.
The majority of radiation-induced thyroid neoplasms are well-differentiated papillary adenocarcinomas, with a lower percentage being of the follicular form. Because of
improved diagnostic procedures and the fact that radiation-induced thyroid cancers do
not usually include the anaplastic and medullary types, the associated mortality rate is
very low (~5/106 per year for all races and both genders). The dose-response data for
thyroid cancer fit a linear pattern. The latency period for benign nodules is approximately 5 to 35 years, and for thyroid malignancies it is approximately 10 to 35 years.
While recent studies of the risk of radiation-induced thyroid cancer have taken different
modeling approaches to estimate the risk (BEIR VII, 2006; NCRP, 2008; EPA, 2011),
the results differ by less than a factor of 2 and all estimates had a range representing the
95% confidence interval that were a factor of 4 or more from their point of estimate.
The relative effectiveness of producing radiogenic thyroid cancer from internally deposited radionuclides has been shown to be approximately 60% to 100% as
effective as external irradiation from x-rays. A comprehensive update of a previous
NCRP report on radiation effects on the thyroid (NCRP, 2008), found that for the
same absorbed dose, I-131 and I-125 were no more than 30% to 60% as effective as
external radiation in causing thyroid cancer with the higher effectiveness applicable
to radioiodine exposure in children. However, the report concluded that other radionuclides including I-123 and Tc-99m were equally effective (per unit thyroid dose)
as external radiation in causing thyroid cancer.
Irradiation of the thyroid may produce other responses, such as hypothyroidism
and thyroiditis. Threshold estimates for adult populations range from 2 Gy for external irradiation to 50 Gy for internal (low-dose-rate) irradiation. Lower threshold
estimates exist for children. Approximately 10% of persons with thyroid doses of 200
to 300 Gy from radioactive iodine will develop symptoms of thyroiditis and/or a sore
throat, and higher doses may result in thyroid ablation.
Due to the considerable ability of the thyroid gland to concentrate iodine, the
thyroid dose from radioiodine is typically 1,000 times greater than in other organs
819
in the body. The relatively large thyroid dose per unit activity administered explains
why thyroid cancer is a major concern following exposure to radioiodine, especially
in children. The Chernobyl nuclear plant accident released large quantities of radioiodine that resulted in a substantial increase in the incidence of thyroid cancer among
children living in heavily contaminated regions. The cows concentrated the radioiodine in their milk by eating contaminated foliage, which was further concentrated in
the thyroid glands of people who consumed the milk. The thyroid dose to children
was higher because they consume more milk and their thyroid glands had higher
concentrations of radioiodine than in adult thyroid glands. Decreased dietary levels
of stable iodine dramatically increase radioiodine uptake by the thyroid. It appears
that dietary iodine deficiency was an important factor in the elevated occurrence
of thyroid cancer observed in children exposed to radioiodine from the Chernobyl
disaster. The increase so far has been almost entirely papillary carcinoma, which is
the most common malignant tumor of the thyroid in adults, adolescents, and children. There has been no increase in thyroid cancer in cohorts made up of adults
at the time of the accident, consistent with the absence of a thyroid cancer excess
among patients treated with I-131 for Graves and other thyroid disorders. In addition, no other cancers have shown an increased incidence thus far, and no increase
has been detected in the incidence of birth defects.
Breast Cancer
For women exposed to ionizing radiation during mammography, the potential for an
increase in the risk of breast cancer is of particular concern. The high prevalence and
morbidity of breast cancer in the female population, and the often differing professional opinions about the risk and benefits of screening mammography only serve
to exacerbate these concerns. Breast cancer is the most commonly diagnosed cancer
among women regardless of race or ethnicity. According to National Cancer Institute,
one out of every eight women (12%) in the U.S, develops breast cancer with approximately 230,000 new cases of invasive breast cancer and 54,000 cases of noninvasive
(in situ) breast cancer expected to be diagnosed in women each year (SEER, 2011). The
lifetime breast cancer mortality risk for women is approximately 2.8% with significant
variation by race and ethnicity (e.g., annual mortality rates for Black and Hispanic
populations are 32.4 and 15.3 per 100,000 respectively). Other etiologic factors in the
risk of breast cancer include age at first full-term pregnancy, family history of breast
cancer, and estrogen levels in the blood. Women who have had no children or only
one child are at greater risk for breast cancer than women who have had two or more
children. In addition, reduced risks are associated with women who conceive earlier
in life and who breast-fed for a longer period of time. Familial history of breast cancer
can increase the risk twofold to fourfold, with the magnitude of the risk increasing as
the age at diagnosis in the family member decreases. As shown previously for Japan
and the United States, there is also considerable racial and geographic variation in
breast cancer risk. Several investigations have suggested that the presence of estrogen,
acting as a promoter, is an important factor in the incidence and latency associated
with spontaneous and radiation-induced breast cancer.
The BEIR VII committee utilized a pooled analysis of data on A-bomb survivors
and medically exposed persons to evaluate the breast cancer risk for low-LET radiation. The data fit a linear dose-response model, with a dose of approximately 800 mGy
required to double the natural incidence of breast cancer. The data from acute and
fractionated (but high dose rate) exposure studies indicate that fractionation of
the dose reduces the risk of breast cancer induced by low-LET radiation. There
is some evidence that protracted exposure to radiation in children reduces the risk of
820
r adiation-induced breast cancer compared with acute or highly fractionated exposures

(Preston, 2002). The latent period ranges from 10 to 40 years, with younger women
having longer latencies. In contrast to leukemia, there is no identifiable window of
expression; therefore the risk seems to continue throughout the life of the exposed
individual. The lifetime attributable risk of developing breast cancer in a population of women with an age distribution similar to that of the U.S. general population is 310/105/0.1 Gy The risk is very age dependent, being approximately 13 times
higher for exposure at age 5 (914/105/0.1 Gy) than at age 50 (70/105/0.1 Gy), (BEIR
VII, 2006). Using the age-specific values in the BEIR VII report, the lifetime attributable risks of developing breast cancer for females receiving a breast dose of 20 mGy
from a chest CT at age 10, 30, or 50 are approximately 0.14%, 0.06%, and 0.014%,
respectively. Use of the tube current modulation (TCM) capability of modern scanners has substantially lowered the dose and subsequent cancer risk, especially for
younger patients. The following example illustrates the benefit of TCM compared
to fixed mA for chest CT in 50-, 30-, and 10-year-old patients. In this example, we
will assume a nominal breast dose of 15 mGy for a chest CT scan of a 50-year-old
woman using fixed mA. In the absence of any other changes except decreasing body
habitus with age, the breast dose might increase to 20 and 25 mGy for similar exams
in the 30- and 10-year-old patients, receptively. As seen in Table 20-17, the increase
in future cancer risk in the 10-year-old is 17 times that of the 50-year-old due to
the combined effects of both increased dose and greater risk per unit dose associated
with her scan. With TCM activated, the dose to 50-year-old women might increase
somewhat but the dose (and associated risk) reductions in the other two age
groups relative to that for the 50-year-old patient is substantial. It can also be appreciated that while the relative reduction in risk can be dramatic as shown in the
example above, the change to the baseline breast cancer risk at the age of exposure
is much less so. Thus, while the difference in dose reduction techniques illustrated
above has only a minor influence on the cancer risk for an individual patient, considering the high utilization of CT in the United States and elsewhere, the overall public
health impact could be quite important. Further discussion of the effect of TCM on
dose reduction can be found in Chapter 10 and in the literature (Angel, 2009a,b).
Improvements in quality assurance and the introduction of digital mammography
have resulted in a substantial reduction in dose to the breast (see Chapter 8). Women
participating in large, controlled mammographic screening studies have been shown
to have a decreased risk of mortality from breast cancer. The American Cancer Society,
the American College of Obstetricians and Gynecologists, and the American College
of Radiology currently recommend annual mammography examination for women
beginning at age 40. We started this section mentioning that there has been concern
TABLE 20-17EFFECT OF AGE AND USE OF AUTO MA ON BREAST DOSE
AND LIFETIME ATTRIBUTABLE CANCER RISK FROM CHEST CT

CT mA
(FIXED/
AUTO)
BREAST
DOSE
(mGy)
LAR BEIR
VII PER 105
PER 0.1 Gy
LIFETIME
RISK AT
AGE
% LAR RISK
LIFETIME
RISK AT AGE
+ % LAR
10
Fixed
Auto
25
10
712
712
12.3
12.3
0.178
0.071
12.478
12.371
17.0
5.1
2.50
30
Fixed
Auto
20
15
253
253
12.3
12.3
0.051
0.038
12.351
12.338
4.8
2.7
1.33
50
Fixed
Auto
15
20
70
70
10.9
10.9
0.011
0.014
10.911
10.914
1.0
1.0
0.75
AGE
RISK
RATIO
AGE/50
RISK RATIO
FIX/AUTO
mA AT AGE
821
over radiation risks associated with mammographic screening examinations. It is

reassuring that the risk of radiation-induced breast cancer decreases substantially with
age at exposure and, despite the models used for risk estimation, there is no consistent
epidemiologic evidence for significant risks for exposures once a woman has passed
the menopausal ages. Further, the exposures for mammographic images are substantially below the levels for which significantly increased risks have been detected.
Noncancer Radiation Effects

Studies of some radiotherapy patient populations have demonstrated increased risks
for diseases other than cancer, particularly cardiovascular disease. There is also evidence of an increase in cardiovascular disease at much lower doses among A-bomb
survivors; however, there is no statistically significant increase at doses below 0.5 Sv.
This evidence notwithstanding, there is no direct evidence for an increased risk of
cardiovascular disease or other noncancer effects at the low doses associated with
typical occupational exposures and diagnostic imaging procedures and thus far, the
data are inadequate to quantify this risk, if it exists.
20.7 Hereditary Effects of Radiation Exposure

Conclusive evidence of the ability of ionizing radiation to produce genetic effects
was first obtained in 1927 with the experimental observations of radiation-induced
genetic effects in fruit flies (Muller, 1927). Extensive laboratory investigations since
that time (primarily large-scale studies in mice (Russell W, et al, 1958)) have led
scientists to conclude that (1) radiation is a mutagenic agent; (2) most mutations
are harmful to the organism; (3) radiation does not produce unique mutations;
(4)chronic radiation exposure produces fewer heritable effects in offspring than the
same dose delivered acutely; and (5) radiation-induced genetic damage can theoretically occur (like cancer) from a single mutation and appears to be linearly related to
dose (i.e., LNT dose-response model). However, it is reassuring that no human study
has demonstrated transgenerational effects following parental exposure. Specifically,
the studies of over 70,000 children, of A-bomb survivors and the children of cancer
survivors treated with radiotherapy find no evidence for increased risks of malformations, neonatal deaths, stillbirths, chromosomal abnormality, or gene changes that
could be related to the exposure to radiation of fathers and mothers (Signorello, 2010;
Winther, 2009). Although genetic effects were initially thought to be the most significant biologic effect of ionizing radiation, it is clear that, for doses associated with
occupational and medical exposure, the risks are small compared with the spontaneous incidence of genetic anomalies and are secondary to their carcinogenic potential.
Epidemiologic Investigations of Radiation-Induced

Hereditary Effects
Epidemiologic investigations have failed to demonstrate radiation-induced hereditary effects, although mutations of human cells in culture have been shown. For a
given exposure, the mutation rates found in the progeny of irradiated humans are
significantly lower than those previously identified in insect populations. The largest population studied is the A-bomb survivors and their progeny. Based on current
risk estimates, failure to detect an increase in radiation-induced mutations in this
population is not surprising considering how few are predicted in comparison to the
spontaneous incidence. Screening of 28 specific protein loci in the blood of 27,000
822
children of A-bomb survivors resulted in only two mutations that might have been
caused by radiation exposure of the parents.
Earlier studies of survivors children to determine whether radiation exposure
caused an increase in sex-linked lethal gene mutations that would have resulted in
increased prenatal death of males or alteration of the gender birth ratio were negative.
Irradiation of human testes has been shown to produce an increase in the incidence
of translocations in spermatogonial stem cells, although no additional chromosomal
aberrations have been detected in children of A-bomb survivors. A large cohort study
of more than 90,000 U.S. radiologic technologists (RTs) examined the risk of childhood cancer (less than 20 years) among more than 100,000 offspring born between
1921 and 1984 to the technologists (Johnson, 2008). Parental occupational radiation
exposure of the testis or ovary prior to conception was estimated from work history
data, dosimetry records, and by estimate from the literature for exposure prior to
1960. Despite the size of the study population, no convincing evidence of an increase
in radiation exposure-related risk of childhood cancer in the offspring of RTs was
found. The result from a number of recent studies evaluating the potential for transgenerational effects in the children of cancer survivors treated with radiotherapy have
also been consistently negative (Signorello, 2010; Tawn, 2011; Winther, 2009).
Estimating Genetic Risk

The genetically significant dose is an index of the presumed genetic impact of
radiation-induced mutation in germ cells in an exposed population. The sensitivity of a
population to radiation-induced genetic damage can be measured by the doubling dose,
defined as the dose required per generation to double the spontaneous mutation rate.
The spontaneous mutation rate is approximately 5 106 per locus and 7 to 15 104
per gamete for chromosomal abnormalities. The doubling dose for humans is estimated to be approximately 1 Gy per generation; however, this represents anextrapo
lationfrom animal data. The BEIR VII committee estimated that an exposure of
10 mGy to the parental generation would cause 30 to 47 additional genetic disorders per
1 million births in the succeeding generation, as a result of increases in the number of
autosomal dominant and (to a much lesser degree) sex-linked dominant mutations.
Variations in exposure to natural background do not contribute significantly to
a populations genetic risk. Even a dose of 100 mGy would only be estimated to
produce about 400 additional genetic disorders per 1 million live births in the first
generation (0.4%/Gy), compared with the normal incidence of approximately 1 in
20 or 5% (some estimates are 1 in 10). Therefore, the 100-mGy dose would cause an
increase in the spontaneous rate of genetic disorders of less than 0.8%.
While the estimates of cancer risk attributable to radiation exposure published
in authoritative reports from the BEIR committees, UNSCEAR, NCRP, and ICRP
have not changed greatly in the past two decades, new data and risk modeling have
resulted in a substantial reduction in the estimated radiation-induced risk of heritable
disease. According to a review by the ICRP, previous estimates of radiation-induced
genetic effects from experiments in mice, and the study of genetic data from A-bomb
survivors likely overestimated the risk of heritable disease from radiation exposure
(ICRP, 2006). Taking into consideration substantial advances in the understanding
of human genetic diseases and the process of germ line mutagenesis after radiation
exposure, the ICRP substantially reduced the proportion of detriment ascribed to the
potential heritable effects of radiation exposure. The ICRP currently estimates the
genetic risks, up to the second generation, to be about 0.2% per Gy. The change in
the ICRP estimates of the relative contribution to the total detriment from cancer and
823
TABLE 20-18DETRIMENT-ADJUSTED NOMINAL RISK
COEFFICIENTS FOR STOCHASTIC EFFECTS AT LOW DOSE RATE

(UNIT, % Sv1)
CANCER
HERITABLE
TOTAL
EXPOSED
1990
2007
1990
2007
1990
2007
All
6.0
5.5
1.3
0.2
7.3
5.7
Adult
4.8
4.1
0.8
0.1
5.6
4.2
heritable effects is shown in Table 20-18. For the population at large, the contribution
of heritable effects to the total detriment from radiation exposure has been reduced
from 23% in 1990 to 3.5% in 2006. This has of course also been reflected in a similar
reduction to the tissues weighting factor for the gonads from 0.2 to 0.08.
Typical diagnostic and occupational radiation exposures, although increasing
the dose to the gonads of those exposed, would not be expected to result in any
significant genetic risk to their progeny. Although delaying conception after therapeutic doses of radiation to reduce the probability of transmission of genetic damage
to offspring is effective, it is not a commonly recommended practice for the relatively
low gonadal doses from most diagnostic imaging procedures.
20.8 Radiation Effects In Utero

Developing organisms are highly dynamic systems that are characterized by rapid
cell proliferation, migration, and differentiation. Thus, the developing embryo is
extremely sensitive to ionizing radiation, as would be expected based on Bergonie
and Tribondeaus early characterization of cellular radiosensitivity. The response after
exposure to ionizing radiation depends on a number of factors including (1) total
dose, (2) dose rate, (3) radiation quality, and (4) the stage of development at the
time of exposure. Together, these factors determine the likelihood of potential consequences (if any) of in utero radiation exposure, including such effects as prenatal or
neonatal death, congenital abnormalities, growth impairment, reduced intelligence,
genetic aberrations, and an increase in future risk of cancer.
Radiation Effects and Gestation

The gestational period can be divided into three stages: a relatively short preimplantation
stage, followed by an extended period of major organogenesis, and finally the fetal
growth stage, during which differentiation is complete and growth occurs. Each of
these stages is characterized by different responses to radiation exposure, owing principally to the relative radiosensitivities of the tissues at the time of exposure.
Preimplantation
The preimplantation stage begins with the union of the sperm and egg and continues
through day 9 in humans, when the zygote becomes embedded in the uterine wall.
During this period, the two pronuclei fuse, cleave, and form the morula and blastula.
The conceptus is very sensitive during the preimplantation stage and susceptibility
to the lethal effects of irradiation is a concern. However, for doses less than 100 mGy,
824
the risks are very low. Embryos exhibit the so-called all-or-nothing response, to radiation exposure at this stage of development, in which, if the exposure is not lethal, the
damaged cells are repaired or replaced to the extent that there is unlikely to be any
additional radiation-induced risk of congenital abnormalities beyond that which would
occur for other reasons. Several factors, including repair capability, lack of cellular differentiation, and the relatively hypoxic state of the embryo, are thought to contribute to
its resistance to radiation-induced abnormalities. During the first few divisions, the cells
are undifferentiated and lack predetermination for a particular organ system. If radiation exposure were to kill some cells at this stage, the remaining cells could continue
the embryonic development without gross malformations because they are indeterminate. However, any misrepaired chromosomal damage at this point may be expressed at
some later time. When cells become specialized and are no longer indeterminate, loss
of even a few cells may lead to anomalies, growth retardation, or prenatal death. The
most sensitive times of exposure in humans are at 12 hours after conception, when the
two pronuclei fuse to the one-cell stage, and again at 30 and 60 hours when the first
two divisions occur. Chromosomal aberrations from radiation exposure at the one-cell
stage could result in loss of a chromosome in subsequent cell divisions that would then
be uniform throughout the embryo. Most chromosomal loss at this early stage is lethal.
Loss of a sex chromosome in female embryos may result in Turner syndrome, although
there is no evidence that this has occurred following radiation exposure.
A woman may not know she is pregnant during the preimplantation period, the
time at which the conceptus is at greatest risk of lethal effects. Animal experiments
have demonstrated an increase in the spontaneous abortion (prenatal death) rate
after doses as low as 50 to 100 mGy delivered during the preimplantation period.
After implantation, doses in excess of 250 mGy are required to induce prenatal death.
The risk of spontaneous abortion following a dose of 10 mGy to the conceptus is
approximately 1%, compared to the naturally occurring rate, which is reported to be
between 30% and 50%. The LD50 for stages from the zygote to expanded blastocysts
is in the range of 1 Gy (ICRP, 2003a).
Organogenesis
Embryonic malformations occur more frequently during the period of major organogenesis (second to eighth week after conception). The initial differentiation of cells to
form certain organ systems typically occurs on a specific gestational day. For example,
neuroblasts (stem cells of the CNS) appear on the 18th gestational day, the forebrain
and eyes begin to form on day 20, and primitive germ cells are evident on day 21.
Each organ system is not at equal risk during the entire period of major organogenesis. In general, the greatest probability of a malformation in a specific organ system
(the so-called critical period) exists when the radiation exposure is received during the
period of peak differentiation of that system. This may not always be the case however, because damage can occur to adjacent tissue, which has a negative effect on a
developing organ system. Some anomalies may have more than one critical period. For
example, cataract formation has been shown to have three critical periods in mice.
The only organ system (in humans or laboratory rodents) that has shown an association between malformations and low-LET radiation doses less than 250 mGy is the
CNS. Embryos exposed early in organogenesis exhibit the greatest intrauterine growth
retardation, presumably because of cell depletion. In utero exposure to doses greater
than 100200 mGy of mixed neutron and gamma radiation from the Hiroshima
atomic bomb resulted in a significant increase in the occurrence of microcephaly.
In general, radiation-induced teratogenic effects are less common in humans than
in animals. This is primarily because, in humans, a smaller fraction of the gestational
9-14
Days Post-conception
15-50
Prenatal Death
825
50-280
Neuropathology
Growth Retardation
Effect
Malformations
Pre
Implantation
Organogenesis
Carcinogenic Risk
Fetal Growth
FIGURE 20-35 Relative sensitivity for radiation-induced effects during different stages of fetal development. (Adapted from Mettler FA, Upton AC. Medical effects of ionizing radiation. 3rd ed. Philadelphia, PA:
Saunders, Co., 2008.)
period is taken up by organogenesis (about 115 of the total, compared with 1 3 for
mice). Nevertheless, development of the CNS in humans takes place over a much
longer gestational interval than in experimental animals, and therefore the CNS is
more likely to be a target for radiation-induced damage. An important distinguishing
characteristic of teratogenic effects in humans are the concomitant effects on the CNS
and/or fetal growth. All cases of human in utero irradiation that have resulted in gross
malformations have been accompanied by CNS abnormalities or growth retardation
or both. The response of each organ to the induction of radiation-induced malformations is unique. Such factors as gestational age; radiation quantity, quality, and dose
rate; oxygen tension; the cell type undergoing differentiation and its relationship to
surrounding tissues; and other factors influence the outcome.
Fetal Growth Stage

The fetal growth stage in humans begins after the end of major organogenesis (day 50) and
continues until term. During this period the occurrence of radiation-induced prenatal death and congenital anomalies is, for the most part, negligible, unless the exposures
are exceptional high and in the therapeutic range. Anomalies of the nervous system
and sense organs are the primary radiation-induced abnormalities observed during this
period, which coincides with their relative growth and development. Much of the damage induced at the fetal growth stage may not be manifested until later in life as behavioral alterations or reduced intelligence (e.g., IQ). Figure 20-35 summarizes the relative
sensitivity for radiation-induced in utero effects during different gestational periods.
Epidemiologic Investigations of Radiation Effects In Utero

Teratogenic Effects
Two groups that have been studied for teratogenic effects of in utero irradiation are
the children of the A-bomb survivors and children whose mothers received medical irradiation (diagnostic and/or therapeutic) during pregnancy. The predominant
effects that were observed included microcephaly and mental and growth retardation. Eye, genital, and skeletal abnormalities occurred less frequently. Excluding the
mentally retarded, individuals exposed in utero at Hiroshima and Nagasaki between
the 8th and 25th week after conception demonstrated poorer IQ scores and school
826
performance than did unexposed children. No such effect was seen in those exposed
before the 8th week or after the 25th week. The decrease in IQ was dose dependent
(~25 points/Gy), with an apparent threshold at approximately 100 mGy below
which, if there was any effect, it was considered negligible. The greatest sensitivity
for radiation-induced mental retardation is seen between the 8th and 15th weeks,
during which the risk for severe mental retardation (SMR) to the fetus is approximately 44% per Gy with a threshold for SMR of approximately 200 to 300 mGy
(ICRP, 2003a).
Microcephaly was observed in children exposed in utero at the time of the
atomic bomb detonations. For those exposed before the 16th week, the incidence
of microcephaly was proportional to the dose, from about 100200 mGy up to 1.5
Gy, above which the decreased occurrence was presumably due to the increase in
fetal mortality. The occurrence of microcephaly increased when exposures occurred
during the first or second trimester and decreased dramatically for exposure in the
third trimester. For exposure prior to 16 weeks, the occurrence of microcephaly for
a fetal dose in the range of 100 to 190 mGy was approximately 11% and increased
to approximately 40% for doses of approximately 1 Gy. Due to differences in the
periods of rapid proliferation and radiosensitivity of the glial and neuronal cells,
microcephaly can appear without SMR between 0 and 7 weeks whereas they
appear together at 8 to 15 weeks and SMR can occur without microcephaly at
16 to 25 weeks.
Similar effects have been reported after in utero exposure during medical irradiation of the mother. Twenty of twenty-eight children irradiated in utero as a result
of pelvic radium or x-ray therapy were mentally retarded, among which sixteen
were also microcephalic (Goldstein, 1929). Other deformities, including abnormal
appendages, hydrocephaly, spina bifida, blindness, cataracts, and microphthalmia
have been observed in humans exposed in utero to high (radiotherapy) doses.
Although each exposure should evaluated individually, the prevailing scientific opinion is that there are thresholds for most congenital abnormalities. Doses
lower than 100 mGy to perhaps 200 mGy are generally thought to carry negligible risk compared with the reported occurrence of congenital anomalies (4% to
6%) in liveborn children. Recommendation of therapeutic abortion for fetal
doses less than 100 mGy, regardless of the gestational age, would not usually be justified. At one time, radiation was widely used to induce therapeutic
abortion in cases where surgery was deemed inadvisable. The standard treatment
was 3.5 to 5 Gy given over 2 days, which typically resulted in fetal death within
1 month. However, in diagnostic imaging, even several abdominal/pelvic CT
scans or dozens of chest CT scans would not expose the fetus to doses in excess
of 100 mGy. In fact, in most instances, fetal irradiation from diagnostic procedures rarely exceeds 50 mGy (see Appendix E-8), which has not been shown to
place the fetus at any significant increased risk for congenital malformation or
growth retardation.
Carcinogenic Effects
According to the U.S. National Cancer Institute, a newborn male and female have
0.25% and 0.22% probability of developing cancer by age 15, respectively (i.e.,
1 in 400 males and 1 in 450 in females, respectively) (NIH, 2009). The majority
of these cancers are leukemia and CNS neoplasms. A correlation between childhood leukemia and solid tumors and in utero diagnostic radiation was reported by
Stewart in 1956 in a retrospective study of childhood cancer in Great Britain (often
referred to as the Oxford Survey of Childhood Cancers [OSCC]; Stewart, 1956).
827
This observation was supported by similar case-control studies reporting a RR of

approximately 1.4 for childhood cancer based on interviewing mothers of children
with and without cancer (UNSCEAR, 1996; Doll and Wakeford, 1997; Wakeford
and Little, 2003). However, no childhood leukemias and only one childhood cancer
were observed among survivors of the atomic bombs who were irradiated in utero.
Absence of a positive finding in the A-bomb survivors could have been predicted on
a statistical basis, owing to the relatively small sample size, but the much higher fetal
dose compared with diagnostic radiography points to an inconsistency in the studies
(1 childhood cancer observed versus 8.8 predicted based on fetal dose). These positive studies have been criticized based on a variety of factors, including the influence
of the preexisting medical conditions for which the examinations were required; the
lack of individual dosimetry and the possibility of recall bias in that mothers with
children who died of cancer may have had more reliable memories of their prenatal
exposures than mothers who did not have such a traumatic experience.
A small increase of approximately 40% in the risk of childhood cancer is associated with acute exposure to diagnostic radiation in utero at doses in the range of 10
mGy, based on results from case-control studies of subjects born between late 1940s
and early 1980s (Doll and Wakeford, 1997). If these estimates are correct, they represent a notable exception to the previously mentioned limit on the general sensitivity of radioepidemiological studies to detect cancer from acute exposures less than
approximately 100-mGy. This may reflect the combined effect of a low background
incidence of, and an increased fetal radiosensitivity to, childhood cancer.
Doll and Wakeford reviewed the scientific evidence and concluded that irradiation of the fetus in utero (particularly in the last trimester) does increase the risk of
childhood cancer; that the increase in risk is produced by doses on the order of 10
mGy, and that the excess risk is approximately 6% per Gy. However, the reported
association with prenatal x-ray and childhood cancer is not uniformly accepted
as causal (ICRP, 2003a). The argument against causality is that the Oxford survey
reported the same risk (odds ratio) for each childhood malignancy, including leukemia, lymphoma, CNS, Wilms tumor, or neuroblastoma. Such a commonality of
risks has never been reported in any radiation study, even among atomic bomb survivors exposed in childhood. The biological implausibility of the case-control studies
is indicated by the high risks seen for Wilms tumor and neuroblastoma, which are
embryonic malignancies and not likely to be induced following pelvimetry examinations occurring shortly before birth. High dose exposures among children and newborns have not been correlated with increases in Wilms tumor or neuroblastoma. In
addition, no cohort study has reported an increase in childhood leukemia following
prenatal x-rays, even a large series of over 30,000 exposed mothers followed in the
United Kingdom. This finding is further supported by animal studies that also failed
to demonstrate an increased risk of leukemia following in utero exposures. Summarizing their view of the increased RR reported in the OSCC, the ICRP stated, Although
the arguments fall short of being definitive because of the combination of biological
and statistical uncertainties involved, they raise a serious question of whether the
great consistency in elevated RRs, including embryonal tumors and lymphomas, may
be due to biases in the OSCC study rather than a causal association. (ICRP, 2003a)
Thus the cancer risk associated with in utero exposure and the often cited risk estimate of 6% per Gy for childhood cancer remain controversial.
This controversy notwithstanding, it is instructive to place the possibly increased risk
of childhood cancer from medical x-rays into perspective by evaluating the magnitude
of the potential risk from common diagnostic imaging procedures. Recall that the baseline risk of a male developing a malignancy through age 15 years is 1/400. If a two view
828
chest x-ray delivered 2 mGy to the conceptus, (and employing the conservative assumption of a causal association between exposure and increased cancer risk at low dose of
6% per Gy) the probability of developing cancer during childhood from the exposure
would be less than 1 in 8.3 million! Even a chest CT, which might deliver a dose to the
conceptus that is on the order of 100 times greater than a chest x-ray (i.e., 200 mGy)
still results in a small additional statistical risk of approximately 1/83,000 (0.0012%) of
developing a childhood cancer. Using the 6%/Gy risk coefficient suggested by Doll, the
doubling dose for childhood malignancies induced by radiation in utero would be on
the order of 35 to 40 mGy. However, current estimates of the risk coefficient, based on
further study of atomic bomb survivors, are on the order of 0.6%/Gy and may be lower
than the risk following exposures in early childhood (Preston 2008).
The lifetime risk of cancer is also believed to be elevated following radiation exposure in utero. ICRP considers the risk to be of the same order of magnitude as the estimated increased risk from radiation exposure during childhood compared to exposures
as adults. The Commissions 2006 report states that it considers that it is prudent to
assume that life-time cancer risk following in utero exposure will be similar to that following
irradiation in early childhood, i.e., at most, about three times that of the population as a whole,
ICRP, 2006. However, the most recent data from the in utero exposed atomic bomb
survivors indicates an increased risk of cancer later in life, it is much lower than the
incidence seen for survivors exposed under the age of 6 years, (Preston 2008). While
there is little doubt about the relative increase in cancer risk (compared with adult
populations) from childhood exposures of some organs and tissues early in life (e.g.,
breast, thyroid) the magnitude of the increase and the applicability of these data for all
radiogenic cancers (and for in utero exposures) remains open to debate. A summary of
fetal effects from low-level radiation exposure is shown in Table 20-19.
U.S. regulatory agencies limit occupational radiation exposure of the fetus to no
more that 5 mSv during the gestational period, provided that the dose rate does not
substantially exceed 0.5 mSv in any 1 month. If twice the maximally allowed dose
were received, the statistical probability of developing cancer during childhood from
the exposure would be approximately 1 in 1,600. Table 20-20 presents a comparison
of the risks of radiation exposure with other risks encountered during pregnancy.
It is clear from these data that, although unnecessary radiation exposure should be
minimized, its risk at levels associated with occupational and diagnostic exposures is
nominal when compared with other potential risk factors.
Risks from In Utero Exposure to Radionuclides

Radiopharmaceuticals may be administered to pregnant women if the diagnostic information to be obtained outweighs the risks associated with the radiation dose. The principal
risk in this regard is the dose to the fetus. Radiopharmaceuticals can be divided into two
broad categories: those that cross the placenta and those that remain on the maternal side
of the circulation. Radiopharmaceuticals that do not cross the placenta irradiate the fetus
by the emission of penetrating radiation (mainly gamma rays and x-rays). Early in the
pregnancy, when the embryo is small and the radiopharmaceutical distribution is fairly
uniform, the dose to the fetus can be approximated by the dose to the ovaries. Estimates
of early-pregnancy fetal doses from commonly used radiopharmaceuticals are listed in
Appendix F-4A. Radiopharmaceuticals that cross the placenta may be distributed in the
body of the fetus or be concentrated locally if the fetal target organ is mature enough. A
classic (and important) example of such a radiopharmaceutical is radioiodine.
Radioiodine rapidly crosses the placenta, and the fetal thyroid begins to concentrate radioiodine around the 11th to 12th week post conception. The ability of
ND
815 (56105)
1625 (112175)
Severe Mental
Retardation (SMR)
No threshold
observed
ND
No threshold
observed
100
600900
200300
ND
250
200
ND
HUMAN STUDIES
0.060.6c
2.5 IQ pts
0.9%
4%
5%10%
ND
ND
ND
ABSOLUTE
INCIDENCEa
Leukemia is the most common type of childhood cancer.

Magnitude of the risk is controversial.
Effects from a dose of 100 mGy or less were statistically

unrecognizable.
At 815 wk 2.5 pts 95% CI (1.83.3)
At 1625 wk 2.1 pts 95% CI (1.2-3.1)
At 1625 wk 95% CI (0%1.8%)
At 815 wk 95% CI (2.2%5.7%)

No increase in absolute incidence was observed for
exposure in the first 7 weeks or after the 25th wk.
MC can appear without severe SMR between 0-7 wks

(see text)
Coincides with period of peak organ system

development
A-bomb survivors who received .200 mGy were

23 cm shorter and 3 kg lighter than controls and had,
on average, head circumferences 1 cm smaller.
If the conceptus survives, it is thought to develop fully,

with no radiation damage.
COMMENTS
Note: IQ, intelligence quotient; ND, no data or inadequate data; CI, confidence interval; wk, weeks; d, days.
a
Absolute incidence is defined as the percentage of exposed fetuses in which an effect is expected to be observed with an equivalent dose of 100 mGy.
b
Organ malformation is defined as malformation of an organ outside the CNS. Data regarding the most sensitive period after conception are from animal studies.
c
The baseline risk of childhood cancer up to age 15 for unexposed fetuses is approximately 0.25% (1 in 400).
Source: Adapted and modified from McCollough, 2007; ICRP, 2003a.
2-term (14-term)
ND
215 (14105)
Microcephaly (MC)
Childhood cancer
250
28 (1456)
Organ
malformationb
815 (56105)
10
18 (856)
Growth retardation
Reduction of IQ
250
100
50100
Postimplantation
01 (08)
Prenatal death
ANIMAL STUDIES
THRESHOLD DOSE AT WHICH AN

EFFECT WAS OBSERVED (mGy)
Preimplantation
weeks (days)
EFFECT
MOST SENSITIVE PERIOD

AFTER CONCEPTION
TABLE 20-19 FETAL EFFECTS FROM LOW-LEVEL RADIATION EXPOSURE

829
830
TABLE 20-20 EFFECT OF RISK FACTORS ON PREGNANCY-OUTCOME

APPROXIMATE NUMBER
OCCURRING FROM
NATURAL CAUSES
RISK FACTOR
EFFECT
ESTIMATED EXCESS
OCCURRENCES FROM RISK
FACTORS
Radiation Risks
Childhood Cancer
Cancer incidence in 3/1,000
children (015)
Radiation dose of 10 0.06/1,000 to 0.6/1,000

mGy received before
birth
Abnormalities
Radiation dose of 10
mGy received during
specific periods after
conception:
Small head size
40/1,000
47 wk
5/1,000
Small head size
40/1,000
811 wk
9/1,000
815 wk
BT
200/1,000
Work in high-risk
occupations
90/1,000
Fetal alcohol
syndrome
12/1,000a
24 drinks/d
100/1,000
Fetal alcohol
syndrome
12/1,000a
More than 4 drinks/d 200/1,000
Fetal alcohol
syndrome
12/1,000a
Chronic alcoholic
(.10 drinks/d)
350/1,000
Perinatal infant
death
23/1,000
Chronic alcoholic
(.10 drinks/d)
170/1,000
Perinatal infant
death
23/1,000
,1 pack/d
5/1,000
Perinatal infant
death
23/1,000
1 pack/d
10/1,000
Mental retardation 4/1,000

Nonradiation Risks
Occupation
Stillbirth or
spontaneous
abortion
Alcohol Consumption
Smoking
There is a naturally occurring syndrome that has the same symptoms as a fetal alcohol syndrome that occurs
in children born to mothers who have not consumed alcohol.
BT, Below threshold for established effects.
Source: Apdapted from U.S. Nuclear Regulatory Commission. Instruction concerning prenatal radiation
exposure. Regulatory Guide 8.13, Rev. 2.
the fetal thyroid to concentrate iodine remains relatively low until the 22nd week,
after which it increases progressively and eventually exceeds that of the mother.
The biologic effect on the fetal thyroid is activity dependent and can result in hypothyroidism or ablation if the dose administered to the mother is high enough. For
I-131, estimates of dose to the fetal thyroid range from 230 to a maximum of 580
mGy/MBq (851 to 2,146 rad/mCi) for gestational ages between 3 and 5 months.
Total body fetal dose ranges from 0.072 mGy/MBq (0.266 rad/mCi) early in pregnancy to a maximum of 0.27 mGy/MBq (1 rad/mCi) near term. Table 20-21 presents
the fetal thyroid dose from various radioiodines and Tc-99m.
831
TABLE 20-21ABSORBED DOSE mGy/MBq (rad/mCi) IN FETAL THYROID
FROM RADIONUCLIDES GIVEN ORALLY TO MOTHER

FETAL AGE (mo)
IODINE-123
IODINE-125
IODINE-131
TECHNETIUM-99m
2.7 (10)
290 (1,073)
230 (851)
0.032 (0.12)
2.6 (10)
240 (888)
260 (962)
6.4 (24)
280 (1,036)
580 (2,146)
6.4 (24)
210 (777)
550 (2,035)
4.1 (15)
160 (592)
390 (1,443)
4.0 (15)
150 (555)
350 (1,295)
2.9 (11)
120 (444)
270 (1,000)
0.15 (0.54)
0.38 (1.40)
Source: Watson EE. Radiation absorbed dose to the human fetal thyroid. In: Watson EE, Schlaske-Stelson A,
eds. 5th International Radiopharmaceutical Dosimetry Symposium. Oak Ridge, TN. May 710, 1991. Oak Ridge,
TN: Oak Ridge Associated Universities, 1992.
Summary
Radiation exposure of the embryo at the preimplantation stage usually leads to an allor-nothing phenomenon (i.e., either fetal death and resorption or normal fetal risk).
During the period of organogenesis, the risk of fetal death decreases substantially,
whereas the risk of congenital malformation coincides with the peak developmental
periods of various organ systems. Exposures in excess of 1 Gy are associated with a
high incidence of CNS abnormalities. During the fetal growth stage in utero, exposure
poses little risk of congenital malformations; however, growth retardation, abnormalities of the nervous system, and the risk of childhood cancer can be increased depending on fetal dose. Growth retardation after in utero fetal doses in excess of 200 mGy
has been demonstrated. Fetal doses from most diagnostic x-ray and nuclear medicine
examinations are typically much lower than 50 mGy (see Appendices E-8 and F-4A)
and have not been demonstrated to produce any significant impact on fetal growth
and development. A number of epidemiologic studies have evaluated the association
between in utero radiation exposure and childhood neoplasms. Although these studies show conflicting results and remain open to varying interpretations, a reasonable
estimate of the excess risk of childhood cancer from in utero irradiation is approximately 0.6% to 6% per Gy. Figure 20-35 and Table 20-19 summarize the relative
incidence of radiation-induced health effects at various stages in fetal development.
While there has been a justifiable concern, both in public opinion and among
health care providers regarding the potential for overutilization of x-rays in medical
imaging, the concern regarding radiation exposure during pregnancy is often disproportionate to the actual risks involved. In one study evaluating perceptions of teratogenic risk with the use of medical x-rays, approximately 6% of primary care physicians
and gynecologists suggested that therapeutic abortion would be appropriate for women
having undergone an abdominal CT scan (Ratnapalan, 2004). Pregnant patients have
also been shown to greatly overestimate the radiation-related risk to the fetus (CohenKerem, 2006). The risk of in utero radiation exposure can be placed in perspective
by considering the probability of birthing a healthy child after a given dose the
conceptus (Table20-22).
Clearly, even a conceptus dose of 100 mGy does not significantly affect the
risks associated with pregnancy. Fetal doses from diagnostic examinations rarely,
if ever, justify therapeutic abortion; however, some patients may receive one or
more high-dose interventional exams or radiation therapy where the fetal doses
832
TABLE 20-22 PROBABILITY OF BIRTHING HEALTHY CHILDREN

DOSE TO
CONCEPTUS
mGya
CHILD WITH
NO MENTAL
RETARDATION (%)
CHILD WITH
NO BIRTH
DEFECT (%)
CHILD WILL
NOT DEVELOP
CANCER (%)b,c
CHILD WILL NOT DEVELOP CANCER

OR HAVE MENTAL RETARDATION
OR BIRTH DEFECT (%)c
99.6
96
99.77
95.40
1.0
99.6
96
99.76
95.39
2.5
99.6
96
99.75
95.38
99.6
96
99.74
95.37
10
99.6
96
99.71
95.34
100
99.6
96
99.17
94.82
Refers to equivalent dose above natural background. Dose assumed to be delivered during the most sensitive
period of gestation (mental retardation: 815 weeks post conception; malformation: 27 weeks post conception).
b
Assumes conservative risk estimates, and it is possible that there is no added radiation-induced cancer risk.
Childhood (015 years) cancer risk from fetal irradiation of 6%/Gy Source: Doll R, Wakeford R. Risk of childhood cancer from fetal irradiation Br J Radiol 1997;70:130139; Wakeford R, Little MP. Risk coefficients for
childhood cancer after intrauterine irradiation: a review. Int J Radiat Biol 2003;79:293309; and ICRP 84 (2000)
Pregnancy and Medical Radiation Vol 30, No 1.
c
Precision displayed is only for the purpose of showing the magnitude of the potential change in risk as a
function of dose and should not be interpreted as a measure of precision with which these outcomes can be
predicted.
Source: Adapted and modified from Wagner LK, Hayman LA. Pregnancy in women radiologists. Radiology
1982;145:559562 and ICRP Publication 84 (ICRP, 2000).
a
are significant. Each case should be evaluated individually and the risks should be
explained to the patient. Many professional organizations including the NCRP, ICRP,
the American College of Radiology (ACR) and the American College of Obstetricians and Gynecologists (ACOG) have issued policy statements regarding the use
of therapeutic abortion and the relatively low risk of diagnostic imaging procedures
resulting in radiation exposure of the fetus. The ACR policy states The interruption of
pregnancy is rarely justified because of radiation risk to the embryo or fetus from a radiologic examination. (ACR, 2005). The ACOG policy, which was revised in 2004, states
Women should be counseled that x-ray exposure from a single diagnostic procedure does
not result in harmful fetal effects. Specifically, exposure to less than 5 rad [50 mGy] has not
been associated with an increase in fetal anomalies or pregnancy loss. (ACOG, 2004).
Every effort should be made to optimize radiation exposure of the patient and
to reduce or avoid fetal exposure whenever possible. However, considering the relatively small risk associated with diagnostic examinations, postponing clinically necessary examinations or scheduling examinations around the patients menstrual cycle to
avoid irradiating a potential conceptus are unwarranted measures. Nevertheless, every
fertile female patient should be asked whether she might be pregnant before diagnostic
examinations or therapeutic procedures using ionizing radiation are performed. If the
patient is pregnant and alternative diagnostic or therapeutic procedures are inappropriate, the risks and benefits of the procedure should be discussed with the patient.
SUGGESTED READINGS
Alliance for Radiation Safety in Pediatric Imaging. Image Gently. http://www.pedrad.org/associations/5364/
ig/. Accessed August 1, 2011.
Amundson S, et al. Low-dose radiation risk assessment. Report of an International Workshop on Low Dose
Radiation Effects held at Columbia University Medical Center, New York, April 34, 2006. Radiat Res
2006;166:561565.
833
Boice JD Jr., Miller RW. Childhood and adult cancer after intrauterine exposure to ionizing radiation.
Teratology 1999;59(4):227233.
Boice JD Jr. Ionizing radiation. In: Schottenfeld D, Fraumeni JF Jr, eds. Cancer epidemiology and prevention.
3rd ed. New York: Oxford University Press, 2006.
Boice JD Jr. Lauriston S. Taylor lecture: Radiation epidemiology: the golden age and future challenges.
Health Phys 2011;100:5976.
Brenner DJ, Doll R, Goodhead DT, et al. Cancer risks attributable to low doses of ionizing radiation:
assessing what we really know. Proc Natl Acad Sci USA 2003;100:1376113766.
Brent RL. Saving lives and changing family histories: appropriate counseling of pregnant women and men
and women of reproductive age, concerning the risk of diagnostic radiation exposures during and
before pregnancy. Am J Obstetr Gynecol 2009;200(1):424.
Dauer LT, Brooks AL, Hoel DG, et al. Review and evaluation of updated research on the health effects
associated with low-dose ionising radiation. Radiat Prot Dosimetry 2010;140(2):103136.
Douple EB, Mabuchi K, Cullings HM, Preston DL, Kodama K, Shimizu Y, Fujiwara S, Shore RE. Long-term
radiation-related health effects in a unique human population: lessons learned from the atomic bomb
survivors of Hiroshima and Nagasaki. Disaster Med Public Health Prep 2011;5(Suppl 1):S122S133.
Fry RJM, Boice JD Jr. Radiation carcinogenesis. In: Souhami RL, Tannock I, Hohenberger P, Horiot JC, eds.
Oxford textbook of oncology. 2nd ed. New York: Oxford Press, 2002:167184.
Hall EJ, Giaccia AJ. Radiobiology for the radiologist. 7th ed. Philadelphia, PA: Lippincott Williams and
Wilkins, 2012.
Hall EJ, Metting N, Puskin J, et al. Low dose radiation epidemiology: what can it tell us? Radiat Res
2009;172:134138.
International Agency for Research on Cancer. IARC monographs on the evaluation of carcinogenic risks to
humans. Vol 75. Ionizing Radiation, Part 1: X- and Gamma (g) Radiation, and Neutrons. Lyon, France:
IARC, 2000.
ICRP Report 90. Biological Effects after Prenatal Irradiation (Embryo and Fetus). Ann ICRP 2003;33(12).
IAEA: Radiation Protection of Patients https://rpop.iaea.org/RPOP/RPoP/Content/index.htm
Joiner M, van der Kogel A. Basic clinical radiobiology. 4th ed. London: Hodder Arnold, 2009.
Mettler FA, Upton AC. Medical effects of ionizing radiation. 3rd ed. Philadelphia, PA: W.B. Saunders Co.,
2008.
National Research Council, Committee on the Biological Effects of Ionizing Radiations, Health Effects of
Exposure to Low Levels of Ionizing Radiation (BEIR VII). Washington, DC: National Academy Press,
2006.
Wagner LK, et al. Radiation bioeffects and management test and syllabus. American College of Radiology
Professional Self-Evaluation and Continuing Education Program, Number 32. Reston, VA: American
College of Radiology, 1991.
Wagner LK, Lester RG, Saldana LR. Exposure of the pregnant patient to diagnostic radiations: a guide to medical
management. 2nd ed. Madison, WI: Medical Physics Publishing, 1997.
West CM, Martin CJ, Sutton DG, et al. 21st L H Gray Conference: the radiobiology/radiation protection
interface. Br J Radiol 2009;82(977):353362.
References
ACOG (2004) ACOG Committee on Obstetric Practice. Guidelines for diagnostic imaging during
pregnancy. ACOG Committee opinion no. 299, September 2004 (replaces no. 158, September 1995).
Obstet Gynecol 2004;104:647651.
Ainsbury EA, et al. Radiation cataractogenesis: A review of recent studies. Radiat Res. 2009; 172:19.
American College of Radiology. ACoR 0405 bylaws. Reston, VA: American College of Radiology, 2005.
American Cancer Society. Cancer facts and figures 2011. Atlanta, GA: American Cancer Society, 2011.
Angel E, Yaghmai N, Jude CM, et al. Monte Carlo simulations to assess the effects of tube current
modulation on breast dose for multidetector CT. Phys Med Biol 2009a;54:497512.
Angel E, Yaghmai N, Jude CM, et al. Monte Carlo simulations to assess the effects of tube current modulation
on breast dose for multidetector CT. Phys Med Biol 2009b54(3):497512; doi: 10.1088/00319155/54/3/003.
Balter S, Hopewell JW, Miller DL, et al. Fluoroscopically guided interventional procedures: a review of
radiation effects on patients skin and hair. Radiology 2010;254(2):326341.
BEIR. Health risks from exposure to low levels of ionizing radiation: BEIR VII, Phase Committee to Assess Health
Risks from Exposure to Low Levels of Ionizing Radiation, Board of Radiation Effects, Research Division on
Earth and Life Studies, National Research Council of the National Academies. National Academy of
Sciences, Washington, DC: National Academies Press, 2006.
834

Boice JD Jr, Hendry JH, Nakamura N, et al. Low-dose-rate epidemiology of high background radiation
areas. Radiat Res 2010;173:849854.
Bernstein JL, Thomas DC, Shore RE, et al. Radiation-induced second primary breast cancer and BRCA1
and BRCA2 mutation carrier status: A report from the WECARE Study. [JNCI In press?]
Brenner D, Elliston C, Hall E, et al. Estimated risks of radiation-induced fatal cancer from pediatric CT.
AJR Am J Roentgenol 2001;176(2):289296.
Brenner DJ, Hall EJ. Computed tomographyan increasing source of radiation exposure. N Engl J Med
2007;357:22772284.
Brenner DJ, Ward JF. Constraints on energy deposition and target size of multiply-damaged sites associated
with DNA double-strand breaks. Int J Radiat Biol 1992;61:737.
Brody AS, Frush DP, Huda W, et al.; American Academy of Pediatrics Section on Radiology. Radiation risk
to children from computed tomography. Pediatrics 2007;120(3):677682.
Chambers C, Fetterly K, Holzer R, et al. Radiation safety program for the cardiac catheterization laboratory.
Catheter Cardiovasc Interv. 2011;77:546556
Clifton DK, Bremner WJ. The effect of testicular x-irradiation on 6872 spermatogenesis in man. A
comparison with the mouse. J Androl 1983;4:387392.
Cohen-Kerem R, Nulman I, Abramow- Newerly M, et al. Diagnostic radiation in pregnancy: Perception
versus true risks. J Obstet Gynaecol Can 2006;28:4348.
Correa CN, Wu X-C, Andrews P, et al., eds. Cancer incidence in five continents, vol VIII Lyon, France: IARC
Scientific Publications No. 155, 2002.
Court Brown WM, Doll R, Hill RB. Incidence of leukaemia after exposure to diagnostic radiation in utero.
Br Med J 1960;2(5212):15391545.
Dekaban AS. Abnormalities in children exposed to x-irradiation during various stages of gestation:
tentative time table of radiation injury to human fetus. J Nucl Med. 1968;9:471477.
De Santis M, Di Gianantonio E, Straface G, et al. Ionizing radiations in pregnancy and teratogenesis. A
review of literature. Reprod Toxicol 2005;20:323329.
Doll R, Wakeford E. Risk of childhood cancer from fetal irradiation. Br J Radiol 1997;70:130139.
EPA. U.S. Environmental Protection Agency. EPA Radiogenic Cancer Risk Models and Projections for the
U.S. Population, EPA 402-R-11-001, April 2011. Fed Reg 2011;76(104).
Fazel R, Krumholz HM, Wang Y, et al. Exposure to low-dose ionizing radiation from medical imaging
procedures. N Engl J Med 2009;361(9):849857.
FDA (1995). U.S. Food and Drug Administration. Radiation-Induced Skin Injuries from Fluoroscopy, http://
www.fda.gov/Radiation-EmittingProducts/ RadiationEmittingProductsandProcedures/MedicalImaging/
MedicalX-Rays/ucm116682.htm (accessed July 11, 2011) (U.S. Food and Drug Administration,
Rockville, Maryland).
Frieben H. Demonstration eines Cancroid des rechten Handruckens, das sich nach langdauernder
Einwirkung von Rontgenstrahlen entwickelt hatte. Fortschr Rontgenstr 1902;6:106111.
Gabriel C. Risk of cataract after exposure to low doses of ionizing radiation: A 20-year prospective cohort
study among US radiologic technologists. Am J Epidemiol. 2008;168(6): 620631.
Goldstein L, Murphy DP. Etiology of ill-health in children born after postconceptional maternal irradiation.
Am J Roentgenol 1929;22:322331.
Goodhead DT. Initial events in the cellular effects of ionizing radiations: Clustered damage in DNA. Int J
Radiat Biol 1994;65:717.
Goodhead DT. Spatial and temporal distribution of energy. Health Phys 1988;55:231240.
Graham SJ, et al. Whole-body pathologic analysis of a patient with thorotrast-in-induced myelodysplasia.
Health Phys 1992;63(1):2026.
Granel F, Barbaud A, Gillet-Terver MN,et al. Chronic radiodermatitis after interventional cardiac
catheterization: Four cases [in French]. Ann Dermatol Venereol. 1998;125:405407.
Hall EJ, Giaccia AJ. Radiobiology for the radiologist. 7th ed. Philadelphia, PA: Lippincott Williams and
Wilkins, 2012.
Henry FA, et al. Fluoroscopy-induced chronic radiation dermatitis: A report of three cases. Dermatol
Online J. 2009;15(1):16.
Hollowell JG Jr, Littlefield LG. Chromosome damage induced by plasma of x-rayed patients: an indirect
effect of x ray. Proc Soc Exp Biol Med 1968;129:240244
Howlader N, Noone AM, Krapcho M, et al., eds. SEER Cancer Statistics Review, 19752008, Bethesda,
MD:National Cancer Institute. http://seer.cancer.gov/csr/1975_2008/, based on November 2010 SEER
data submission, posted to the SEER web site, 2011.
International Commission on Radiological Protection. 1990 recommendations of the international commission
on radiological protection. ICRP Publication 60. Elsevier, New York: Ann ICRP 21(13), 1991.
International Commission on Radiological Protection. Genetic susceptibility to cancer. ICRP publications 79.
Tarrytown, NY:Ann. ICRP Permagon Press, 1999.
835
International Commission on Radiological Protection. Pregnancy and medical radiation. ICRP Publication
84. Elsevier, New York: Ann. ICRP 30(1), 2000.
International Commission on Radiological Protection. Biological effects after prenatal irradiation (embryo and
fetus). ICRP Publication 90. Ann ICRP 2003a;33(1/2).
International Commission on Radiological Protection. Relative Biological Effectiveness (RBE), Quality
Factor (Q), and Radiation Weighting Factor (wR). ICRP Publication 92. Ann ICRP 2003b;33(4).
International Commission on Radiological Protection. Low-dose extrapolation of radiation-related cancer
risk. ICRP Publication 99. Ann ICRP 2006;35:1140.
International Commission on Radiological Protection. The 2007 Recommendations of the International
Commission on Radiological Protection. ICRP Publication 103. Ann ICRP 2007a;37:1332.
International Commission on Radiological Protection. Radiological Protection in Medicine, ICRP Publication
105. Ann ICRP 2007b;37(6).
International Commission on Radiological Protection. ICRP reference 4825-3093-1464. Statement on
Tissue Reactions Approved by the Commission on April 21, 2011. 2011a.
International Commission on Radiological Protection. Early and late effects of radiation in normal 16 tissues
and organs: threshold doses for tissue reactions and other non-cancer effects of radiation in a radiation
protection context ICRP Draft Report For Consultation: reference 4844-6029-7736, January 20, 2011.
2011b.
Johnson KJ, Alexander BH, Doody MM, et al. Childhood cancer in the offspring born in 19211984 to
US radiologic technologists. Br J Cancer 2008;99:545550.
Kuppermann N, Holmes JF, Dayan PS; et al, Pediatric Emergency Care Applied Research Network
(PECARN). Identification of children at very low risk of clinically-important brain injuries after head
trauma: a prospective cohort study. Lancet 2009;374(9696):11601170.
Little JB. Genomic instability and bystander effects: a historical perspective. Oncogene 2003;22:
69786987.
Lbrich M, Cooper PK, Rydberg B. Non-random distribution of DNA double-strand breaks induced by
particle irradiation. Int J Radiat Biol 1996;70:493.
Lushbaugh CC, Ricks RC. Some cytokinetic and histopathologic considerations of irradiated male and
female gonadal tissues. Front Radiat Ther Oncol 1972;6:228248.
McCollough CH, Schueler BA, Atwell TD, et al. Radiation exposure and pregnancy: When should we be
concerned? Radiographics 2007;27:909917.
Mettler FA, Upton AC. Medical effects of ionizing radiation. 3rd ed. Philadelphia, PA: W.B. Saunders Co.,
2008.
Muller HJ, Artificial transmutation of the gene. Science 1927;66:8487
Nair RR, Rajan B, Akiba S, et al. Background radiation and cancer incidence in Kerala, India-Karanagappally
cohort study. Health Phys. 2009;96:5566.
National Cancer Institute;. PDQ Cancer Information Summary. Genetics of Breast and Ovarian Cancer
(PDQ)Health Professional. Bethesda, MD Date last modified 06/23/2011. Available at: http://
www.cancer.gov/cancertopics/pdq/genetics/breast-and-ovarian/healthprofessional, 2011. Accessed
07/15/2011.
National Council on Radiation Protection and Measurements. Limitation of exposure to ionizing radiation.
NCRP Report No. 116. National Council on Radiation Protection and Measurements, Bethesda, MD,
1993.
National Council on Radiation Protection and Measurements. Radiation protection guidance for activities in
low-earth orbit. NCRP Report No. 132. Bethesda, MD: National Council on Radiation Protection and
Measurements, 2000.
National Council on Radiation Protection and Measurements. Extrapolation of radiation-induced cancer
risks from nonhuman experimental systems to humans. NCRP Report No. 150. Bethesda, MD: National
Council on Radiation Protection and Measurements, 2005.
National Council on Radiation Protection and Measurements. Risk to thyroid from ionizing radiation.
NCRP Report No. 159. Bethesda, MD: National Council on Radiation Protection and Measurements,
2008.
National Council on Radiation Protection and Measurements. Radiation dose management for fluoroscopicallyguided interventional procedures. NCRP Report No. 168. Bethesda, MD: National Council on Radiation
Protection, 2010.
NIH. Probability of Developing or Dying of Cancer Software, Version 6.4.1. Statistical Research and
Applications Branch, National Cancer Institute, 2009. http://srab.cancer.gov/devcan.
NTP. Report on Carcinogens, Twelfth Edition. Research Triangle Park, NC: U.S. Department of Health and
Human Services, Public Health Service, National Toxicology Program, 2011.
Preston DL, Cullings H, Suyama A, Funamoto S, Nishi N, Soda M, Mabuchi K, Kodama K, Kasagi F, Shore
RE. Solid cancer incidence in atomic bomb survivors exposed in utero or as young children. J Natl
Cancer Inst 2008;100(6):428436.
836

Preston DL, Mattsson A, Holmberg E, Shore R, Hildreth NG, Boice JD Jr. Radiation effects on breast cancer
risk: a pooled analysis of eight cohorts. Radiat Res 2002;158:220235.
Preston DL, Pierce DA, Shimizu Y, et al. Effect of recent changes in atomic bomb survivor dosimetry on
cancer mortality., risk estimates. Radiat Res 2004;162:377389.
Preston DL, Shimizu Y, Pierce DA, et al. Studies of mortality of atomic bomb survivors. Report 13: Solid
cancer and noncancer disease mortality: 19501997. Radiat Res 2003;160:381407.
Preston DL, et al. Solid cancer incidence in atomic bomb survivors: 19581998. Radiat Res 2007;168:1.
Russell WL, Russell LB, Kelly EM. Radiation dose rate and mutation frequency. Science 1958;128:
15461550.
Sasaki MS. Primary damage and fixation of chromosomal DNA as probed by monochromatic soft x-rays
and low-energy neutrons. In: Fielden EM, ONeil P, eds. The Early Effects of Radiation on DNA. Vol.
H54. NATO ASI Series, Berlin: Springer-Verlag, 1991:369384.
Sasaki MS, Kobayashi K, Hieda K, et al. Induction of chromosome aberrations in human lymphocytes
by monochromatic x-rays of quantum energy between 4.8 and 14.6 keV. Int J Radiat Biol
1989;56:975988.
Schull WJ, Neel JV. Atomic bomb exposure and the pregnancies of biologically related parents. A
prospective study of the genetic effects of ionizing radiation in man. Am J Public Health Nations Health
1959;49(12):16211629.
Shadley JD, Afzal V, Wolff S. Characterization of the adaptive response to ionizing radiation induced by
low doses of x rays to human lymphocytes. Radiat Res 1987;111:511517.
Shope TB. Radiation-induced skin injuries from fluoroscopy. RadioGraphics 1996;16:11951199.
Shrieve, Loeffler. Human radiation injury. Philadelphia, PA: Wolters Kluwer Health/Lippincott Williams &
Wilkins, 2011.
Signorello LB, Mulvihill JJ, Green DM, Munro HM, Stovall M, Weathers, RE, Mertens AC, Whitton
JA, Robison LL, Boice JD Jr. Stillbirth and neonatal death in relation to radiation exposure before
conception: a retrospective cohort study. Lancet 2010;376:624630.
Stewart A, Webb J, Hewitt D. A survey of childhood malignancies. Brit Med J 1958;30:14951508.
Tawn EJ, Rees GS, Leith C, Winther JF, Curwen GB, Stovall M, Olsen JH, Rechnitzer C, Schroeder H,
Guldberg P, Boice JD Jr. Germline minisatellite mutations in survivors of childhood and young adult
cancer treated with radiation. Int J Radiat Biol 2011;87:330340.
United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). UNSCEAR 1996.
Sources and Effects of Ionizing Radiation. 1996 Report to the General Assembly with Scientific Annex.
New York: United Nations.
United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). UNSCEAR 2006
Report to the General Assembly with Scientific Annexes, Effects of Ionizing Radiation. Vol 1: Report
and Annexes A and B, 2008. Vol 2: Annexes C, D and E, 2009, 2011, New York: United Nations.
Upton AC, Odell TT Jr., Sniffen EP. Influence of age at time of irradiation on induction of leukemia and
ovarian tumors in RF mice. Proc Soc Exp Biol Med 1960;104:769772.
Wakeford R, Little MP. Risk coefficients for childhood cancer after intrauterine irradiation: a review. Int J
Radiat Biol 2003;79:293309.
Ward JF. DNA damage produced by ionizing radiation in mammalian cells: Identities, mechanisms of
formation, and repairability. Prog Nucleic Acid Res Mol Biol 1988;35:95125.
Ward JF. The complexity of DNA damage: Relevance to biological consequences. Int J Radiat Biol
1994;66:427432.
Winther JF, Boice JD Jr, Fredericksen K, Bautz A, Mulvihill JJ, Stovall M, Olsen JH. Radiotherapy for
childhood cancer and risk for congenital malformations in offspring: a population-based cohort study.
Clin Genet 2009;75:5056.
Young R, Kerr G. Report of the joint US-Japan working group. Reassessment of the atomic bomb
radiation dosimetry for Hiroshima and Nagasaki: dosimetry system DS02. Radiation Effects Research
Foundation, 2005.
Chapter
21
Radiation Protection
It is incumbent upon all individuals who use radiation in medicine to strive for an
optimal compromise between its clinical utility and the radiation doses to patients,
staff, and the public. Federal and state governments and even some large municipalities have agencies that promulgate regulations regarding the safe use of radiation and radioactive material. Another goal of radiation protection programs is to
ensure compliance with these regulations. To a large degree, the success of radiation protection programs depends on the development of procedures for the safe
use of radiation and radioactive material and on the education of staff about radiation safety principles, the risks associated with radiation exposure and contamination, and the procedures for safe use. This chapter discusses the application of
radiation protection principles (also known as health physics) in diagnostic x-ray
and nuclear imaging, image-guided interventional procedures, and therapy with
unsealed radioactive material.
21.1 Sources of Exposure to Ionizing Radiation

According to the National Council on Radiation Protection and Measurements
(NCRP) Report No. 160, the average annual per capita effective dose, exclusive of
doses to patients from external beam radiation therapy, from exposure to ionizing
radiation in the United States in 2006 was approximately 6.2 millisievert (mSv).
Approximately half of this, about 3.1 mSv, was from naturally occurring sources,
whereas about 48%, 3.0 mSv, was from medical exposure of patients. Only about
2%, 0.14 mSv, was from other sources, such as consumer products and activities and
occupational exposure. These averages apply to the entire population of the United
States. Doses to individuals from these sources vary considerably with a variety of
factors discussed below.
Ubiquitous Background Exposure

Naturally occurring sources of radiation include (1) cosmic rays, (2) cosmogenic
radionuclides, and (3) primordial radionuclides and their radioactive decay products.
Cosmic radiation includes both the primary extraterrestrial radiation that strikes the
Earths atmosphere and the secondary radiations produced by the interaction of primary cosmic rays with the atmosphere. Primary cosmic rays predominantly consist
of extremely penetrating high-energy (mean energy ~10 GeV) particulate radiation,
approximately 80% of which is high-energy protons. Almost all primary cosmic radiation collides with our atmosphere before reaching the ground, producing showers of
secondary particulate radiations (e.g., electrons and muons) and electromagnetic radiation. The average per capita effective dose from cosmic radiation is approximately
0.33 mSv per year or approximately 11% of natural background radiation. However,
the range of individual exposures is considerable. The majority of the population of
837
838
the United States is exposed to cosmic radiation near sea level where the outdoor
effective dose rate is approximately 0.3 mSv per year. However, smaller populations
receive much more than this amount (e.g., Colorado Springs, Colorado, at 1,840 m,
~0.82 mSv per year). Exposures increase with altitude, approximately doubling every
1,500 m, as there is less atmosphere to attenuate the cosmic radiation. Cosmic radiation is also greater at the Earths magnetic poles than at the equator, as charged particles encountered by the Earths magnetic field are forced to travel along the field lines
to either the North or South Pole. Structures provide some protection from cosmic
radiation; the indoor effective dose rate is approximately 20% lower than outdoors.
Some of the secondary cosmic ray particles collide with stable atmospheric nuclei
producing cosmogenic radionuclides (e.g., 147 N n,p 146 C) . Although many cosmogenic radionuclides are produced, they contribute very little (~0.01 mSv per year
or less than 1%) to natural background radiation. The majority of the effective dose
caused by cosmogenic radionuclides is from carbon 14.
The radioactive materials that have been present on the Earth since its formation
are called primordial radionuclides. Primordial radionuclides with physical half-lives
comparable to the age of the Earth (~4.5 billion years) and their radioactive decay
products are the largest sources of terrestrial radiation exposure. The population radiation dose from primordial radionuclides is the result of external radiation exposure,
inhalation, and incorporation of radionuclides in the body. Primordial radionuclides
with half-lives less than 108 years have decayed to undetectable levels since their formation, whereas those with half-lives greater than 1010 years do not significantly contribute to background radiation levels because of their long physical half-lives (i.e.,
slow rates of decay). Most radionuclides with atomic numbers greater than lead decay
to stable isotopes of lead through a series of radionuclide decays called decay chains.
The radionuclides in these decay chains have half-lives ranging from seconds to many
thousands of years. Other primordial radionuclides, such as potassium 40 (K-40, T
= 1.28 109 years), decay directly to stable nuclides. The decay chains of uranium
238 (U-238), T = 4.51 3 109 years (uranium series), and thorium 232 (Th-232),
T = 1.41 3 1010 years (thorium series), produce several dozen radionuclides that
together with K-40 are responsible for most of the external terrestrial average effective
dose of 0.21 mSv per year or approximately 7% of natural background. Individuals
may receive much higher or lower exposures than the average, depending on the
local concentrations of terrestrial radionuclides. The range in the United States is
approximately 0.1 to 0.4 mSv per year. There are a few regions of the world where terrestrial radionuclides are highly concentrated. For example, as discussed in Chapter
20, monazite sand deposits, containing high concentrations of radionuclides from the
Th-232 decay series, are found along certain beaches in India. The external radiation
levels on these black sands range up to 70 mGy per year (Nair et al., 2009), which is
more than 300 times the average level from terrestrial sources in the United States.
The short-lived alpha particle-emitting decay products of radon 222 (Rn-222) are
believed to be the most significant source of exposure from the inhalation of naturally
occurring radionuclides. Radon-222, a noble gas, is produced in the U-238 decay
chain by the decay of radium 226 (Ra-226). Rn-222 decays by alpha emission, with a
half-life of 3.8 days, to polonium 218 (Po-218), followed by several other alpha and
beta decays, eventually leading to stable lead-206 (Pb-206). When the short-lived
daughters of radon are inhaled, most of the dose is deposited in the tracheobronchial
region of the lung. Radon concentrations in the environment vary widely. There are
both seasonal and diurnal variations in radon concentrations. Radon gas emanates
primarily from the soil in proportion to the quantity of natural uranium deposits; its
dispersion can be restricted by structures, producing much higher indoor air concentrations than found outdoors in the same area. Radon gas dissolved in domestic water
Chapter 21 Radiation Protection
839
supplies can be released into the air within a home during water usage, particularly
when the water is from a well or another ground water source. Weatherproofing of
homes and offices and other energy conservation measures typically decrease outside
air ventilation, resulting in higher indoor radon concentrations.
The radiation from exposure to Rn-222 and its daughters in the United States
results in an average effective dose of approximately 2.1 mSv per year or approximately 68% of natural background. The dose from inhalation of radon decay products is primarily to the bronchial epithelium. In order to convert the absorbed dose
from the alpha particles to effective dose, a tissue weighting factor (wT) of 0.08 is
applied along with a radiation weighting factor (wR) of 20 to account for increased
risk from exposure to high LET radiation.
The average indoor air concentration of Rn-222 in homes in the United States
is approximately 46 Bq/m3 (1.24 pCi/L); however, levels can exceed 2.75 kBq/m3
(75 pCi/L) in poorly ventilated structures with high concentrations of U-238 in
the soil. Outdoor air concentrations are approximately three times lower, 15 Bq/m3
(0.41 pCi/L). The U.S. Environmental Protection Agency recommends taking action
to reduce radon levels in homes exceeding 147 Bq/m3 (4 pCi/L), whereas other countries have somewhat different limits (e.g., the United Kingdom and Canada have
higher action levels, 200 Bq/m3 [5.4 pCi/L]). Although Rn-222 accounts for about
two thirds of the natural radiation effective dose, it can be easily measured and exposures can be reduced when necessary.
The third largest source of natural background radiation is from ingestion of
food and water containing primordial radionuclides (and their decay products), of
which K-40 is the most significant. K-40 is a naturally occurring isotope of potassium (~0.01%). Skeletal muscle has the highest concentration of potassium in the
body. K-40 produces an average effective dose of approximately 0.15 mSv per year,
or approximately 5% of natural background. Th-232 and U-238 and their decay
products are found in food and water and result in an average annual effective dose
of approximately 0.13 mSv.
Long-lived radionuclides released to or created in the environment from the
atmospheric testing of nuclear weapons (450 detonations between 1945 and 1980)
consist mainly of carbon 14, tritium (H-3), cesium 134 and 137 (Cs-134, Cs-137),
strontium 90, plutonium, and transplutonium elements. A large fraction of these
radionuclides have since decayed and/or have become progressively less available for
biologic uptake. Today, the dose from this radioactive material is minimal and results
in an average annual effective dose less than 10 Sv (NCRP, 1987).
Medical Exposure of Patients

The single greatest controllable source of radiation exposure in the US population is
from medical imaging. The majority of the exposure is from x-ray imaging (primarily from diagnostic radiology), with a smaller contribution from nuclear medicine.
Doses from individual medical imaging procedures are summarized in Appendices
E and F. In 2006, the medical use of radiation, not including radiation therapy, produced an average annual effective dose to the population of the United States of
approximately 3 mSv, which is about 97% of the total from artificial radiation sources
and nearly half of the average annual effective dose from all sources (NCRP, 2009).
There have been significant changes in medical imaging technology and its uses since
1980, when the estimated average annual effective dose equivalent from the medical exposure of patients was about 0.53 mSv (NCRP, 1987). The two largest factors
that increased the average effective dose from 1980 until 2006 are the increased
utilization of computed tomography (CT) and nuclear medicine imaging procedures
840
Conventional
radiography &
fluoroscopy
(68%)
Medical Exposure
of Patients (early 1980s)
Computed
tomography
(49%)
Medical Exposure
of Patients (2006)
Computed
tomography
(3%)
Nuclear
medicine
(26%)
Interventional
fluoroscopy
(3%)
Average U.S. Per

Capita Dose
1980: 0.53 mSv
2006: 3.0 mSv
Conventional
radiography &
fluoroscopy
(11%)
Increase
of 566%
Nuclear
medicine
(26%)
Interventional
fluoroscopy
(14%)
FIGURE 21-1 Average annual per capita effective doses in the United States from medical imaging procedures, showing the large increase from 1980 to 2006. (NCRP, 1987; NCRP, 2009.)
(Fig. 21-1). While CT and nuclear medicine procedures collectively account for only
approximately 22% of all medical imaging procedures using ionizing radiation, they
deliver approximately 75% of the collective effective dose. The CT and nuclear medicine procedures that contribute the most to this dose are CT of the abdomen and
of the pelvis and Tc-99m and Tl-201 myocardial perfusion imaging, which together
account for more than half of the effective dose to the population from medical
imaging. Conversely, conventional radiographic and noninterventional fluoroscopic
imaging procedures, which account for approximately 74% of the imaging procedures, result in only 11% of the collective effective dose (Table 21-1). Data from the
State of Michigan from 2002 to 2006 indicate that there has been an increase in the
entrance skin exposure of 67% for a PA chest projection and 28% for an AP lumbar spine projection for digital imaging when compared with screen-film imaging
TABLE 21-1RADIATION EXPOSURE TO US POPULATION IN 2006MEDICAL
EXPOSURES (EXCLUDING RADIATION THERAPY)

NUMBER OF
PROCEDURES
(MILLIONS)
%
Computed Tomography
Nuclear Medicine
COLLECTIVE
EFFECTIVE DOSE
(person Sv)
%
17
22%
5
18
440,000
67
49
75%
26
231,000
EFFECTIVE DOSE
(mSv)
1.5
0.8
Interventional Fluoroscopy
17
128,000
Conventional Radiography
and Fluoroscopy
293
74
99,000
11
0.3
TOTALS
395
100
898,000
100
,3
Data from NCRP (2009).
0.4
841
(NCRP, 2009). Studies have shown that the per-image doses to patients when photostimulable phosphor (computed radiography [CR]) image receptors are used are on
the order of twice the doses when modern 400-speed rare-earth screen-film image
receptors are used (Compagnone et al., 2006; Seibert et al., 1996) and it is likely that
this is the cause of the increase. However, digital radiography equipment with much
more detection efficient thin-film-transistor flat-panel arrays is now replacing many
CR systems, and this equipment can produce images with lower patient doses than
CR or even state of the art film-screen systems (Compagnone et al., 2006; Williams
et al., 2007). If this technology is used effectively, the per procedure and population
doses from radiography may decline significantly.
Consumer Products and Activities

This category includes a variety of sources, most of which are consumer products. The
largest contribution in this category is from tobacco products. Two alpha-emitting radionuclides Pb-210 and Po-210 (which occur naturally from the decay of Ra-226) have
been measured in both tobacco leaves and cigarette smoke. It has been estimated that
a one-pack-a-day smoker increases his or her annual effective dose by approximately
0.36 mSv. The NCRP estimates the average annual effective dose to an exposed individual is 0.3 mSv and, based on a smoking population of 45 million in the United States,
this results in an average annual effective dose to the entire population of 0.045 mSv,
which is about 35% of the effective dose from all consumer products and activities.
Many building materials contain radioisotopes of uranium, thorium, radium,
and potassium. These primordial radionuclides and their decay products are found
in higher concentration in such materials as brick, concrete, and granite, and thus
structures built from these materials will cause higher exposures than structures constructed primarily from wood. The average annual per capita effective dose is estimated to be approximately 0.035 mSv from this source.
Air travel can substantially add to an individuals cosmic ray exposure. For example, a 5-h transcontinental flight in a commercial jet aircraft will result in an equivalent dose of approximately 0.025 mSv. The average annual per capita effective dose to
passengers from commercial air travel is estimated to be approximately 0.034 mSv.
There are many other less important sources of enhanced natural radiation exposure, such as mining and agricultural activities (primarily from fertilizers containing
members of the uranium and thorium decay series and K-40); combustible fuels,
including coal and natural gas (radon); and consumer products, including smoke
alarms (americium-241), gas lantern mantles (thorium), and dental prostheses, certain ceramics, and optical lenses (uranium). These sources contribute less than 12%
of the average annual effective dose from consumer products and activities.
Occupational and Other Sources of Exposure

Occupational exposures are received by some people employed in medicine, including veterinary medicine; by aircraft crew in commercial aviation; by workers in some
industrial and commercial activities; by workers in the commercial nuclear power
industry; by workers in some educational and research institutions; and by some individuals in the military, some in governmental agencies, and some in U.S. Department
of Energy facilities. Since most individuals are not occupationally exposed to radiation
and the majority of those who are exposed typically receive fairly low annual doses,
the contribution to the average annual per capita effective dose to the population from
occupational exposure is very low, 0.005 mSv (less than 0.1%). However, among those
occupationally exposed to radiation, the average annual effective dose was 1.1 mSv.
842
In 2006, medical personnel, whose occupational exposures were monitored, received

an average annual effective dose of approximately 0.75 mSv (NCRP, 2009). This average is somewhat lower than might be expected because the doses to many staff members are quite low (e.g., radiology supervisors and radiologists who perform few if any
fluoroscopic procedures). For example, occupational doses to radiologists not routinely
performing interventional procedures have been declining in recent years and an average
annual effective dose of approximately 0.1 to 0.2 mSv is common (Linet et al., 2010).
Similarly, full-time technologists working in large hospitals performing mostly CT and
radiography examinations will typically have annual effective doses of approximately
0.5 to 1 mSv. However, medical staff involved in fluoroscopically guided interventional
procedures will typically have much higher occupational exposures. The actual doses
received by the staff will depend on a number of factors including their roles in the procedures (i.e., performing or assisting), the number of procedures performed, the type and
difficulty (which determines the lengths) of the cases, as well as the availability and use
of radiation protection devices and techniques. Annual doses recorded by the dosimeters
worn at the collar (outside the lead apron) in the range of 5 to 15 mSv are typical for personnel routinely performing these procedures. In reality, these are only partial body exposures (i.e., to the head and extremities) because the use of radiation-attenuating aprons
greatly reduces the exposure to most of the body. Adjusting their measured exposures
to account for the shielding provided is discussed in the following section on dosimetry.
After making adjustments to account for shielding, the effective dose is typically reduced
by a factor of 3 or more compared to that recorded on the collar dosimeter.
For most routine nuclear medicine procedures, the effective doses to the technologists
are typically less than 1 Sv per procedure, putting typical annual exposures in the range
2 to 3 mSv. However for positron emission tomography (PET) procedures, an effective
dose of 1 to 5 Sv per procedure has been reported (Guillet, 2005). Nuclear medicine
technologists whose routine workload also includes dose preparation and imaging of
patients for PET may have annual effective doses in the range of 10 to 15 mSv.
Airline crew members are estimated to receive an additional average annual effective dose of approximately 3.1 mSv; some receive effective doses of more than twice
this value. It is interesting to note that the average annual effective dose of airline
crew exceeds the annual effective doses of many diagnostic radiology personnel.
The contribution to the annual effective dose of members of the public (those not
working in the industry) from activities related to commercial nuclear power production is minimal, approximately 0.5 Sv. Population radiation exposure from nuclear
power production occurs during all phases of the fuel cycle, including uranium mining and processing, uranium enrichment, manufacturing of uranium fuel, reactor
operations, and radioactive waste disposal.
Summary
The average annual effective dose to the US population from all radiation sources is
obtained by dividing the annual collective effective dose by the size of the US population. The result in 2006 was approximately 6.2 mSv per year or approximately 17 Sv
per day for all people in the United States from all sources, exclusive of radiation therapy. It is interesting to note that, although nuclear power, fallout from atomic weapons,
and a variety of consumer products receive considerable attention in the popular media,
they in fact cause only a small fraction of the average population exposure to radiation.
The average annual effective dose is somewhat misleading for categories such
as the medical use of radiation, in that many people in the population are neither
occupationally nor medically exposed, yet are included in the average. Figure 21-2
843
Radon & thoron

(37%)
T
o
t
a
l
6.2
mSv
per
yr
B
A
C
K
3.1
G mSv/yr
R
50%
O
U
N
D
0.
Terrestrial (3%)
yr
0.3
v/
mS
Internal (5%)
r Cosmic Rays (5%)
0.3 mSv/y
1.5 mSv/yr
yr
Sv/
Computed
tomography
(24%)
5m
0.7
M
E
100% D
3.0
I mSv/yr
C
48%
A
L
/y
Sv
2.3 mSv/yr
0.1 mSv/yr Consumer Products

& Other (2%)
0.33
mSv
/yr Conventional
radiography/
0.4
4m
fluoroscopy
Sv
(5%)
/yr
Interventional
fluoroscopy
(7%)
Nuclear
medicine (12%)
FIGURE 21-2 Percent contributions of various sources of exposure to the annual collective effective dose
(1,870,000 person-Sv) and the average annual effective dose per person in the US population (6.2 mSv) in
2006. Percentages have been rounded to the nearest 1%. (Adapted from NCRP 2006.)
presents a summary of the average annual effective dose for the US population from
the radiation sources previously discussed.
21.2 Personnel Dosimetry

The radiation exposures of some people must be monitored for both safety and regulatory purposes. Such assessments may need to be made over periods of several minutes
to several months. There are three main types of individual radiation recording devices
called personnel dosimeters used in diagnostic radiology and nuclear medicine: (1) film
badges, (2) dosimeters using storage phosphors (e.g., thermoluminescent dosimeters
[TLDs]), and (3) pocket dosimeters, each with its own advantages and disadvantages.
Ideally, one would like to have a single personnel dosimeter capable of meeting all
of the dosimetry needs in medical imaging. The ideal dosimeter would instantaneously
respond, distinguish among different types of radiation, and accurately measure the
dose equivalent from all forms of ionizing radiation with energies from several keV to
MeV, independent of the angle of incidence. In addition, the dosimeter would be small,
lightweight, rugged, easy to use, inexpensive, and unaffected by environmental conditions (e.g., temperature, humidity, pressure) and nonionizing radiation sources. Unfortunately, no such dosimeter exists; however, most of these characteristics can be satisfied
to some degree by selecting the dosimeter best suited for a particular application.
Film Badges
A film badge consists of a small sealed packet of radiation sensitive film, similar
to dental x-ray film, placed inside a special plastic holder that can be clipped to
clothing. Radiation striking the emulsion causes a darkening of the developed film
844
(see Chapter 7). The amount of darkening increases with the absorbed dose to the
film emulsion and is measured with a densitometer. The film emulsion contains
grains of silver bromide, resulting in a higher effective atomic number than tissue;
therefore, the dose to the film is not equal to the dose to tissue. However, with the
selective use of several metal filters over the film (typically lead, copper, and aluminum), the relative optical densities of the film underneath the metal filters can be
used to identify the approximate energy range of the radiation and to calculate the
dose to soft tissue. Film badges typically have an area where the film is not covered
by a metal filter or plastic and thus is directly exposed to the radiation. This open
window is used to detect medium and high-energy beta radiation that would otherwise be attenuated (Fig. 21-3).
Most film badges can record doses from about 100 Sv to 15 Sv (10 mrem to
1,500 rem) for photons and from 500 Sv to 10 Sv (50 mrem to 1,000 rem) for beta
radiation. The film in the badge is usually replaced monthly and sent to the commercial supplier for processing. The developed film is usually kept by the vendor,
providing a permanent record of radiation exposure.
Film badges are small, lightweight, inexpensive, and easy to use. However, exposure to excessive moisture or temperature can damage the film emulsion, making
dose estimates difficult or impossible.
90
Sr (beta)
B C
Cs
662 keV
118 keV
73 keV
53 keV
35 keV
20 keV
137
FIGURE 21-3 A film pack (A) consists of a light opaque envelope (B) containing the film (C). The film pack is placed
in the plastic film badge (D) sandwiched between two sets of Teflon (E), lead (F), copper (G), and aluminum (H)
filters. Film badges typically have an area where the film pack is not covered by a filter or the plastic of the badge
and thus is directly exposed to the radiation. This open window area (I) is used to detect medium and high-energy
beta radiation that would otherwise be attenuated. The relative darkening of the developed film (filter pattern)
provides a crude but useful assessment of the energy of the radiation. The diagram shows typical filter patterns from
exposure to a high-energy beta emitter (Sr-90), a high-energy gamma emitter (Cs-137), and x-rays with effective
energies from 20 to 118 keV. (Film badge and filter patterns courtesy of Global Dosimetry Solutions, Irvine, CA.)
845
Thermoluminescent and Optically Stimulated Luminescent

Dosimeters
Some dosimeters contain storage phosphors in which a fraction of the electrons,
raised to excited states by ionizing radiation, become trapped in excited states. When
these trapped electrons are released, either by heating or by exposure to light, they
fall to lower energy states with the emission of light. The amount of light emitted can
be measured and indicates the radiation dose received by the phosphor material.
TLDs, discussed in Chapter 17, are excellent personnel and environmental dosimeters; however, they are somewhat more expensive than film badges. The most commonly used TLD material for personnel dosimetry is lithium fluoride (LiF). LiF TLDs
have a wide dose response range of 100 Sv to 10 Sv and are reusable (Fig.21-4). These
dosimeters can be used over a long time interval (up to 6 months if necessary) before
being returned to the vendor for analysis. The energy response is 0.766 to 5 MeV (Emax)
for beta radiation and 20 keV to 6 MeV for x-ray and gamma-ray radiation.
Another advantage of LiF TLDs is that their effective atomic number is close to
that of the tissue; therefore, the dose to the LiF chip is close to the tissue dose over a
wide energy range. TLDs do not provide a permanent record, because heating the chip
to read the exposure removes the deposited energy. TLDs are routinely used in nuclear
medicine as extremity dosimeters; a finger ring containing a chip of LiF worn on the
hand is expected to receive the highest exposure during radiopharmaceutical preparation and administration. Figure 21-5 shows a finger ring dosimeter and a LiF chip.
FIGURE 21-4 TLD dosimeter with four LiF TLDs. The filters in this dosimeter are made of mylar (7 mg/cm2),
copper (300 mg/cm2), and polypropylene plastic (1,000 mg/cm2), representing the specified depths for determination of dose to the skin (shallow dose) at a depth of 0.007 cm, lens of the eye at a depth of 0.3 cm,
and deep dose at a depth of 1.0 cm, respectively. (Technical specifications from Global Dosimetry Solutions
Irvine, CA.)
846
Dosimeters using optically stimulated luminance (OSL) have recently become

commercially available as an alternative to TLDs. The principle of OSL is similar
to that of TLDs, except that the release of trapped electrons and light emission are
stimulated by laser light instead of by heat. Crystalline aluminum oxide activated
with carbon (Al2O3:C) is commonly used. Like LiF TLDs, these OSL dosimeters have
a broad dose response range, and are capable of detecting doses as low as 10 Sv. As
in film dosimeters, the Al2O3 has a higher effective atomic number than soft tissue
and so an OSL dosimeter has filters over the sheet of OSL material that are used to
estimate dose to soft tissue, as in film badges. However, OSL dosimeters have certain
advantages over TLDs in that they can be reread several times and an image of the
filter pattern can be produced to differentiate between static (i.e., instantaneous)
and dynamic (i.e., normal) exposure. TLDs or OSL dosimeters are the dosimeters of
choice when longer dose assessment intervals (e.g., quarterly) are required.
Direct Ion Storage Dosimeters

Direct ion storage dosimeters are a relatively new technology in which a nonvolatile
analog memory cell, surrounded by a gas-filled ion chamber, is used to record radiation
exposure (Fig. 21-6). The initial interactions of the x-ray and gamma-ray photons occur
in the wall material, and secondary electrons ionize the gas of the chamber. The positive
ions are collected on a central negative electrode resulting in a reduction in voltage that
is proportional to the dose received by the dosimeter. The dose recorded by the dosimeter can be read at any time by connecting it to the USB port of any computer with
Internet access. Analysis of the deep, shallow, and lens of the eye dose is accomplished
by a computer algorithm that takes into consideration the typical photon energy spectrum experience of the user. The advantages of this technology include a broad dose
and photon energy response range (0.01 mSv to 5 Sv and 5 keV to 6 MeV), unlimited
real-time dose readings by the user without the need for a special reader, online management of dosimeter assignment and dosimetry reports, and elimination of the periodic distribution and collection of dosimeters as well as the delay and cost associated
with returning the dosimeters for processing by the dosimetry vendor. Disadvantages
include initial cost of the dosimeters, more costly replacement of lost dosimeters, and
the need for users to upload dosimetry information periodically. In addition, the current version of this dosimeter cannot be used to measure exposure to beta radiation.
FIGURE 21-5 A small chip of

LiF (right) is sealed in a finger ring
(underneath the identification label).
In nuclear medicine, the ring is worn
with the LiF chip on the palmar surface
such that the chip would be facing a
radiation source held in the hand.
847
+ + +
+ + +
Air or Gas
Direct Ion Storage
Technical Specifications
Lower Limit of
Detection
0.01 mSv
Useful Dose Range
0.01 mSv to 5 Sv
Photon Energy
Response
5 keV to 6 MeV
FIGURE 21-6 Direct ion storage dosimeter. (Photos courtesy of Mirion Technologies San Ramon, CA, www.
mirion.com.)
Practical Aspects of Dosimeter Use

Nearly every medical facility obtains nonself-reading dosimeters, whether film
badges, TLD dosimeters, and/or OSL dosimeters, from a commercial vendor monthly
or quarterly. One or more control dosimeters are shipped with each batch. At the
beginning of a wear period, typically at the beginning of a month, the new dosimeters are issued to staff and the used dosimeters from the previous wear period are
collected. The used dosimeters are returned to the dosimeter vendor for reading. At
least one control dosimeter from the same batch is included in the shipment. Control
dosimeters are stored in an area away from radiation sources. The vendor subtracts
the reading from the control dosimeter from the readings of the dosimeters that were
used. An exposure report is received from the vendor in about 2 weeks. However,
reporting of unusual exposures or exposures over regulatory limits is expedited. The
dosimetry report lists the shallow dose, corresponding to the skin dose, the eye
dose corresponding to the dose to the lens of the eye and the deep dose, corresponding to penetrating radiations. Most vendors post dosimetry results on password-secured Web sites.
Placement of Dosimeters on the Body

A dosimeter is typically worn on the part of the torso that is expected to receive the
largest radiation exposure or is most sensitive to radiation damage. Most radiologists,
x-ray technologists, and nuclear medicine technologists wear a dosimeter at waist
or shirt-pocket level. During fluoroscopy, a dosimeter is typically placed at collar
level in front of the lead apron to measure the dose to the thyroid and lens of the eye
848
because most of the body is shielded from exposure. Alternatively, a dosimeter can
be placed at the collar level in front of the radiation-protective apron, and a second
dosimeter can be worn on the torso underneath the apron. A pregnant radiation
worker typically wears an additional dosimeter at waist level (behind the lead apron,
if worn) to assess the fetal dose.
Estimating Effective Dose and Effective Dose Equivalent for

Staff Wearing Protective Aprons
When protective aprons, which shield the torso and upper legs, are worn during
diagnostic and interventional x-ray imaging procedures, the effective dose and effective dose equivalent can be estimated by using methods that take the protection from
the shielding into account. When a single dosimeter is worn at collar level outside the
apron, the NCRP (NCRP, 1995) recommends that effective dose equivalent (HE) be
calculated from the dose recorded by the collar dosimeter (HN) using Equation21-1:

HE = 0.18 HN
[21-1]
When a dosimeter is worn at collar level outside the protective apron and another
dosimeter is worn underneath the apron at the waist or chest, the recommended
method for calculating HE is to use HN and the dose (HW) recorded by the dosimeter
worn under the lead apron in Equation 21-2:

HE = 1.5 HW + 0.04 HN
[21-2]
Similar equations for estimating the effective dose can be found in NCRP Report No.
122 (NCRP, 1995).
Pocket Dosimeters
The major disadvantage to film, thermoluminescent, and OSL dosimeters is that
the accumulated dose is not immediately displayed. Pocket dosimeters measure
radiation exposure and can be read immediately. The analog version of the pocket
dosimeter is the pocket ion chamber. It utilizes a quartz fiber suspended within an
air-filled chamber on a wire frame, on which an electrical charge is placed. The fiber
is bent away from the frame because of coulombic repulsion and is visible through
an optical lens system upon which an exposure scale is superimposed. When radiation strikes the detector, ion pairs produced in the air partially neutralize the charge,
thereby reducing coulombic repulsion and allowing the fiber to move closer to the
wire frame. This movement is seen as an up-range excursion of the hairline fiber on
the exposure scale (Fig. 21-7).
Pocket ion chambers can typically detect photons of energies greater than 20 keV.
Pocket ion chambers commonly display the quantity exposure (defined in Chapter 3).
Pocket ion chambers are available in a variety of ranges; the most commonly used models measure exposures from 0 to 200 mR or 0 to 5 R. These devices are small (the size
of a pen) and easy to use; however, they may produce erroneous readings if bumped or
dropped and, although reusable, do not provide a permanent record of exposure.
Digital pocket dosimeters can be used in place of pocket ion chambers. These
dosimeters use solid-state electronics and either Geiger-Mueller (GM) tubes or radiation-sensitive semiconductor diodes to measure and display radiation dose in a range
from approximately 10 Sv to 100 mSv. Figure 21-8 shows analog and digital pocket
dosimeters.
849
Cap with window

Supporting insulator
Wire element (Frame)
Quartz fiber
Electrode
Case
Electroscope
Lenses
Gamma or X-Ray
(lonization)
Dosage scale
Eyepiece
MILLIROENTGEN
0
0
50
100
150
200
20 40 60 80 100 120 140 160 180 200
FIGURE 21-7 Cross-sectional diagram of an analog pocket ion chamber.
Pocket dosimeters can be utilized when high doses are expected, such as during cardiac catheterization or manipulation of large quantities of radioactivity. Table21-2 summarizes the characteristics of the various personnel monitoring devices discussed above.
Problems with Dosimetry

Common problems associated with dosimetry include dosimeters being left in radiation fields when not worn, contamination of a dosimeter itself with radioactive material, lost and damaged dosimeters, and people not wearing dosimeters when working
with radiation sources. If a dosimeter is positioned so that the body is between it and
the radiation source, attenuation will cause a significant underestimation of the true
FIGURE 21-8 Examples of analog and digital pocket dosimeters.
Pocket ion chamber courtesy of
Arrow-Tech, Inc. Rolla, ND, and electronic personnel dosimeter courtesy
of Cone Instruments, Solon, OH.
850
TABLE 21-2 SUMMARY OF PERSONNEL MONITORING METHODS

METHOD
MEASURES
USEFUL RANGE
(X-AND GAMMA RAY)
PERMANENT
RECORD
Film badge
Beta;
gamma
and x-ray
0.115,000 mSv
(beta)
0.510,000 mSva
Yes
Routine personnel monitoring;

most common in diagnostic
radiology and nuclear
medicine
TLD
Beta;
gamma
and x-ray
0.01106 mSva
No
Becoming more common but

still more expensive than film;
used for phantom and patient
dosimetry
OSL
Beta;
gamma
and x-ray
0.01106 mSva
Nob
Advantage over TLD includes

the ability to reread the
dosimeters and distinguish
between dynamic and static
exposures
Pocket
dosimeter
Gamma and
x-ray
Analog Digital
No
Special monitoring
(e.g., cardiac cath); permits
direct (i.e., real-time) reading
of exposure
00.2 R 0100 mSv

00.5 R
05 R
USES AND COMMENTS
Multiply mSv by 100 to obtain mrem.

OSL dosimeters are typically retained and can be reread by the manufacturer for approximately 1 year.
OSL, optically stimulated luminance; TLD, thermoluminescent dosimeter.
exposure. Most personnel do not remain in constant geometry with respect to the
radiation sources they use. Consequently, the dosimeter measurements are usually
representative of the individuals average exposure. For example, if a dosimeter is
worn properly and the radiation field is multidirectional or the wearers orientation
toward it is random, then the mean exposure over a period of time will tend to be a
good approximation (10% to 20%) of the individuals true exposure.
21.3 Radiation Detection Equipment in Radiation Safety

A variety of portable radiation detection instruments, the characteristics of which
are optimized for specific applications, are used in radiology and nuclear medicine.
The portable GM survey meter and portable ionization chamber survey meter satisfy
most of the requirements for radiation protection measurements in nuclear medicine.
X-ray machine evaluations require specialized ion chamber instruments capable of
recording exposure, exposure rates, and exposure durations. All portable radiation
detection instruments should be calibrated at least annually. A small radioactive
check source can be used to verify an instruments response to radiation.
Geiger-Mueller Survey Instruments

GM survey instruments are used to detect the presence and provide semiquantitative
estimates of the intensities of radiation fields. Measurements from GM survey meters
typically are in units of counts per minute (cpm) rather than mR/h, because the GM
detector does not duplicate the conditions under which exposure is defined. In addition,
the relationship between count rate and exposure rate with most GM probes is a complicated function of photon energy. If a GM survey meter is calibrated to indicate exposure
851
rate (most commonly performed using a sealed source containing Cs-137 [662 keV
gamma rays]), one should refer to the detectors energy response curve before making
quantitative measurements of photons whose energies significantly differ from the energy
for which it was calibrated. However, with specialized energycompensated probes, GM
survey instruments can provide approximate measurements of exposure rate (typically
in mR/h) over a wide range of photon energies, although with reduced sensitivity. The
theory of operation of GM survey instruments was presented in Chapter 17.
A common application of GM survey meters is to perform surveys for radioactive contamination in nuclear medicine. A survey meter coupled to a thin window
(~1.5 to 2 mg/cm2), large surface area GM probe (called a pancake probe) is ideally
suited for contamination surveys (see Fig. 17-7). Thin window probes can detect
alpha (greater than 3 MeV), beta (greater than 45 keV), and x- and gamma (greater
than 6 keV) radiations.
These detectors are extremely sensitive to charged particulate radiations with sufficient energy to penetrate the window, but are less sensitive to x- and gamma radiations.
These detectors will easily detect natural background radiation (on the order of ~50
to 100 cpm at sea level). These instruments have long dead times resulting in significant count losses at high count rates. For example, a typical dead time of 100 s will
result in, approximately, a 20% loss at 100,000 cpm. Some GM survey instruments will
saturate in high-radiation fields and read zero, which, if unrecognized, could result in
significant overexposures. Portable GM survey instruments are best suited for low-level
contamination surveys and should not be used in high-radiation fields or when accurate measurements of exposure rate are required unless specialized energycompensated probes or other techniques are used to account for these inherent limitations.
Portable Ionization Chamber Survey Meters

Portable ionization chamber survey meters are used when accurate measurements of
radiation exposure rates are required (see Figure 17-6). These ionization chambers
approximate the conditions under which the roentgen is defined (see Chapter 3).
They have a wide variety of applications including assessment of radiation fields near
brachytherapy or radionuclide therapy patients, and surveys of radioactive material
packages. The principles of operation of ion chambers are discussed in Chapter 17.
The ion chambers of some survey meters are filled with ambient air, whereas
others have sealed ion chambers. Some of those with sealed chambers are pressurized to increase the sensitivity. ionization chamber measurements are influenced
by photon energy and exposure rate and ambient air chambers are also affected by
temperature and atmospheric pressure. However, these limitations are not very significant for conditions typically encountered in medical imaging. For example, a
typical portable ion chamber survey meter will experience only approximately 5%
loss for exposure rates approaching 10 R/h; specialized detectors can measure much
higher exposure rates. Most portable ionization chamber survey meters respond
slowly to rapidly changing radiation exposure rates. These instruments must be
allowed to warm up and stabilize before accurate measurements can be obtained.
Some are also affected by orientation and strong magnetic fields (e.g., from magnetic resonance imaging [MRI] scanners). Ionization chamber survey meters are typically capable of measuring exposure rates between 1 mR/h and 50 R/h for photons
of energies greater than approximately 20 keV. Some ionization chambers have a
cover over one end of the detector, which serves as a buildup cap to establish electronic equilibrium for accurate measurement of higher-energy x- and gamma rays;
it can be removed to improve the accuracy when measuring low-energy x- and
852
gamma rays. Removing the cap also permits assessment of the contribution of beta
particles to the radiation field. The slow response time and limited sensitivity of these
detectors preclude their use as low-level contamination survey instruments or to
locate a lost low-activity radiation source.
21.4 Fundamental Principles and Methods of Exposure
Control
Principles of Justification, Optimization, and Limitation

The International Commission on Radiological Protection (ICRP) has formulated a
set of principles that apply to the practice of radiation protection (ICRP, 2007):
The
principle of justification: Any decision that alters the radiation exposure situation, for example, by introducing a new radiation source or by reducing existing
exposure, should do more good than harm, that is, yield an individual or societal
benefit that is higher than the detriment it causes.
The principle of optimization of protection: Optimization of protection should
ensure the selection of the best protection option under the prevailing circumstances, that is, maximizing the margin of good over harm. Thus, optimization
involves keeping exposures as low as reasonably achievable (ALARA), taking into
account economic and societal factors.
The principle of limitation of maximum doses: In planned situations, the total dose
to any individual from all the regulated sources should not exceed the appropriate
regulatory limits.
The ICRP recommends that the last principle, limitation of maximal doses, not apply
to medical exposure of patients or emergency situations. However, the other two
principles do apply to medical exposure of patients and emergency situations.
Methods of Exposure Control

There are four principal methods by which radiation exposures to persons can be
minimized: (1) reducing time of exposure, (2) increasing distance, (3) using shielding, and (4) controlling contamination by radioactive material. Although these methods are widely used in radiation protection programs, their application to medical
imaging is addressed below.
Time
Although it is obvious that reducing the time spent near a radiation source will reduce
ones radiation exposure, techniques to minimize time in a radiation field are not
always recognized or practiced. First, not all sources of radiation produce constant
exposure rates. Diagnostic x-ray machines typically produce high exposure rates during brief time intervals. For example, a typical chest x-ray produces an entrance skin
exposure of approximately 20 mR in less than 1/20 of a second, an exposure rate of
1,440 R/h. In this case, exposure is minimized by not activating the x-ray tube when
staff are near the radiation source. Nuclear medicine procedures, however, typically
produce lower exposure rates for extended periods of time. The time spent near a
radiation source can be minimized by having a thorough understanding of the tasks
to be performed and the appropriate equipment to complete them in a safe and
timely manner. Similarly, radiation exposure to staff and patients can be reduced during fluoroscopy if the operator is proficient in the procedure to be performed.
853
Distance
The exposure rate from a source of radiation decreases with increasing distance from
the source, even in the absence of an attenuating material. In the case of a point source
of radiation (i.e., a source whose physical dimensions are much less than the distance
from which it is being measured), the exposure rate decreases as the distance from the
source is squared. This principle is called the inverse square law and is the result of the
geometric relationship between the surface area (A) and the radius (r) of a sphere: A
= 4r2. Thus, if one considers an isotropic point radiation source at the center of the
sphere, the surface area over which the radiation is distributed increases as the square
of the distance from the source (i.e., the radius). If the exposure rate from a point
source is X1 at distance d1, at another distance d2 the exposure rate X2 will be
2
X 2 = X1 (d1 / d2 )
[21-3]
For example, if the exposure rate at 20 cm from a source is 90 mR/h, doubling the
distance will reduce the exposure by ()2 = to 22.5 mR/h; increasing the distance
to 60 cm decreases the exposure by ( 1 3 )2 = 1 9 to 10 mR/h (Fig. 21-9).
This relationship is only valid for point sources (i.e., sources whose dimensions
are small with respect to the distances d1 and d2). Thus, this relationship would not
be valid near (e.g., less than 1 m from) a patient injected with radioactive material. In
this case, the exposure rate decreases less rapidly than 1/(distance)2.
This rapid change in radiation intensity with distance is familiar to all who have
held their hand over a heat source, such as a stove top or candle flame. There is only
a fairly narrow range over which the infrared (thermal, nonionizing) radiation felt
is comfortable. A little closer and the heat is intolerable; a little further away and
the heat is barely perceptible. Infrared radiation, like all electromagnetic radiation,
follows the inverse square law.
Source
Inverse Square Law
K2 = K1 (d1/d2)2
20 cm
900 mGy/hr
40 cm
K40 = 900 mGy/hr (20/40)2 = 225 mGy/hr
60 cm
K60 = 900 mGy/hr (20/60)2 = 100 mGy/hr
FIGURE 21-9 The inverse square law, showing the decreasing air kerma rate with distance from a point
source.
854
Scattered radiation from a patient, x-ray tabletop, or shield is also a source of personnel radiation exposure. For diagnostic energy x-rays, a good rule of thumb is that
at 1 m from a patient at 90 degrees to the incident beam, the radiation intensity is
approximately 0.1% to 0.15% (0.001 to 0.0015) of the intensity of the beam incident
upon the patient for a 400 cm2 x-ray field area on the patient (typical field area for fluoroscopy). All personnel should stand as far away from the patient as practicable during
x-ray imaging procedures and behind a shielded barrier or out of the room, whenever
possible. The NCRP recommends that personnel should stand at least 2 m from the
x-ray tube and the patient during radiography with mobile equipment (NCRP, 1989).
Because it is often not practical to shield the technologist from the radiation emitted from the patient during a nuclear medicine exam (see Shielding), distance is the
primary dose reduction technique. The majority of the nuclear medicine technologists annual whole-body radiation dose is received during patient imaging. Imaging
rooms should be designed to allow a large distance between the imaging table and
the computer terminal where the technologist spends most of the time during image
acquisition and processing.
Shielding
Shielding is used in diagnostic radiology and nuclear medicine to reduce exposures of
patients, staff, and the public. The decision to utilize shielding, and its type, thickness,
and location for a particular application, are functions of the photon energy, intensity
and geometry of the radiation sources, exposure rate goals at various locations, and other
factors. The principles of attenuation of x- and gamma rays are reviewed in Chapter 3.
Shielding may be installed in a wall, floor, or ceiling of a room; this is commonly
done for rooms containing x-ray imaging machines, PET/CT machines, and sometimes rooms housing patients administered large activities. It may be in the walls of a
cabinet used to store radioactive sources. It may be installed around a work area, such
as the dose preparation area in a nuclear medicine radiopharmacy. It may be around
individual sources, such as an x-ray tube or a vial or syringe of radioactive material in
nuclear medicine. It is incorporated behind the image receptors of fluoroscopes and
radiographic machines and in the gantries of CT devices. It may be worn as protective
aprons by people performing fluoroscopy. It may also be in movable barriers, such as
the freestanding and ceiling-mounted shields in fluoroscopic procedure rooms.
The x-rays from nearly all medical x-ray machines used for diagnostic and interventional imaging have photon energies that do not exceed 140 keV and the average
energies are much less. The gamma rays emitted by the radionuclides used in nuclear
medicine mostly have energies less than 365 keV, although some of these radionuclides emit higher-energy gamma rays of low abundance. The annihilation photons
used in PET have energies of 511 keV. For photons of these energies, use of a high
atomic number material for shielding, due to photoelectric absorption, requires less
mass of shielding than a material of lower atomic number. The use of a material of
high density permits a thinner shield. Thus, lead is commonly used for shielding,
because of its very high atomic number, density, and reasonable cost. In general, placing shielding closer to a source of radiation does not reduce the thickness needed, but
does reduce the mass of shielding necessary by reducing the area of the shielding.
21.5 Structural Shielding of Imaging Facilities

The purpose of radiation shielding of rooms containing x-ray machines is to limit
radiation exposures of employees and members of the public to acceptable levels.
Several factors must be considered when determining the amount and type of radiation
855
shielding. Personnel exposures may not exceed limits established by regulatory agencies. Furthermore, personnel radiation exposures must be kept ALARA.
Methods and technical information for the design of shielding for diagnostic and
interventional x-ray rooms are found in NCRP Report No. 147, Structural Shielding
Design for Medical X-ray Imaging Facilities (NCRP, 2004). The recommended quantity
for shielding design calculations is air kerma (K), with the unit of Gy; typical annual
amounts of air kerma in occupied areas are commonly expressed in mGy. The recommended radiation protection quantity for the limitation of exposure of people to
sources of ionizing radiation is effective dose (E), defined as the sum of the weighted
equivalent doses to specific organs or tissues (the equivalent dose to each organ or
tissue being multiplied by a corresponding tissue weighting factor, wT), expressed
in Sv (see Chapter 3 for definition); for protection purposes, typical levels of E are
expressed in mSv.
Areas to be protected by shielding are designated as controlled and uncontrolled
areas; a controlled area is an area to which access is controlled for the purpose
of radiation protection and in which the occupational exposure of personnel to
radiation is under the supervision of a person responsible for radiation protection.
Controlled areas, such as procedure rooms and control booths, are usually in the
immediate vicinity of the x-ray and nuclear medicine imaging devices. The workers in controlled areas are usually radiologic technologists, nurses, radiologists, and
other physicians trained in the use of ionizing radiation and whose radiation exposures are typically individually monitored. Uncontrolled areas for radiation protection purposes are most other areas in the hospital or clinic, such as offices adjacent
to x-ray rooms.
Shielding design goals, P, are amounts of air kerma delivered over a specified
time at a stated reference point that are used in the design and evaluation of barriers
constructed for the protection of employees and members of the general public from
a medical x-ray or radionuclide imaging source or sources. Shielding design goals are
stated in terms of K (mGy) at a reference point beyond a protective barrier (e.g., 0.3 m
for a wall, a conservative assumption of the distance of closest approach). Because of
conservative assumptions, achieving the design goals will ensure that the respective
annual recommended values for E are not exceeded. The relationship between E and
K is complex, and depends on several factors, including x-ray energy spectrum and
the posture (e.g., standing or sitting) of the exposed individual. Because E cannot be
directly measured, it is impractical to use it for a shielding design goal, and therefore,
shielding design goals P are stated in terms of K.
There are different shielding design goals for controlled and uncontrolled areas.
Radiation workers, typically employees, have significant potential for exposure to
radiation in the course of their jobs, and as a result are subject to routine monitoring by personal dosimeters. On the other hand, many people in uncontrolled areas
have not voluntarily chosen to be irradiated and may not be aware that they are
being irradiated. NCRP Report No. 147 recommends that the shielding design goal
P for controlled areas be 5 mGy per year, and that for uncontrolled areas be 1 mGy
per year. These are equivalent to shielding design goals, P, of 0.l mGy per week for
controlled areas and 0.02 mGy per week for uncontrolled areas.
There are also air-kerma design goals for stored radiographic film and loaded
film-screen cassettes to avoid film fogging. A shielding design goal, P, less than 0.1 mGy
is recommended for the period in which radiographic film is stored. Since loaded screenfilm cassettes and CR cassettes awaiting use are more sensitive to radiation, a P not to
exceed 0.5 Gy for the period of storage (on the order of a few days) is recommended.
These shielding design methods are based upon conservative assumptions that
will result in the actual air kerma transmitted through each barrier being much less
856
than the applicable shielding design goal. These assumptions include (1) neglecting
the attenuation of the primary x-ray beam by the patient (the patient typically attenuates the x-ray beam by a factor of 10 to 100); (2) assuming perpendicular incidence
of the radiation on the barrier, which has the greatest transmission through the barrier; (3) ignoring the presence of other attenuating materials in the path of the radiation; (4) assuming a large x-ray beam field size for scattered radiation levels; and (5)
assuming high occupancy factors for uncontrolled areas.
Shielding designed by these methods will keep the effective doses or effective
dose equivalents received by workers in these areas much less than a tenth of the current occupational dose limits in the United States, will keep the dose to an embryo or
fetus of a pregnant worker much less than 5 mGy over the duration of gestation, and
will keep the effective doses to members of the public and employees, who are not
considered radiation workers, less than 1 mSv per year.
Sources of Exposure
The sources of exposure that must be shielded in a diagnostic or interventional x-ray
room are primary radiation, scattered radiation, and leakage radiation (Fig. 21-10).
Scatter and leakage radiation are together called secondary or stray radiation. Primary
radiation, also called the useful beam, is the radiation passing through the open area
defined by the collimator of the x-ray source. The amount of primary radiation depends
on the output of the x-ray tube (determined by the kV, mGy/mAs, and mAs) per
examination, the average number of examinations performed during a week, the fraction of time the x-ray beam is directed toward any particular barrier, the distance to the
point to be protected, and the presence (or absence) of a primary barrier built into the
imaging equipment. Scattered radiation arises from the interaction of the useful beam
with the patient, causing a portion of the primary x-rays to be redirected. For radiation
protection purposes scatter is considered as a separate radiation source with essentially
the same photon energy spectrum (and penetrability) as the primary beam. In general,
the exposure from scattered radiation at 1 m from the patient is approximately 0.1% to
0.15% of the incident exposure to the patient for typical diagnostic x-ray energies with
a 20 cm 20 cm (400 cm2) field area. The scattered radiation is proportional to the field
size, and can be calculated as a fraction of the reference field area. For CT applications,
Primary Radiation
Scattered Radiation
Leakage Radiation
FIGURE 21-10 The sources of exposure in a diagnostic x-ray room. A. Primary radiation emanating from the
focal spot. B. Scattered radiation emanating from the patient. C. Leakage radiation emanating from the x-ray
tube housing (other than the collimated primary radiation).
857
the rectangular collimation over a smaller volume will have a distinct scattered radiation
distribution, and is considered separately, as discussed below. Leakage is the radiation that emanates from the x-ray tube housing other than the useful beam. Because
leakage radiation passes through the shielding of the housing, its effective energy is very
high (only the highest energy photons are transmitted). The exposure due to leakage
radiation is limited by FDA regulations to 0.87 mGy/h (100 mR/h) at 1 m from the tube
housing when the x-ray tube is operated at the maximum allowable continuous tube
current (usually 3 to 5 mA) at the maximum rated kV, typically 150 kV.
The primary and secondary radiation exposure of an individual in an adjacent
area to be protected depends primarily on (1) the amount of radiation produced by
the source; (2) the distance between the patient and the radiation source; (3) the
amount of time a given individual spends in an adjacent area; (4) the amount of
protective shielding between the source of radiation and the individual; and (5) the
distance between the source of radiation and the individual.
Types of Medical X-ray Imaging Facilities

General purpose radiographic installations produce intermittent radiographic exposures using tube potentials of 50 to 150 kV, with the x-ray beam directed toward the
patient and the image receptor. Depending on the type of procedures, a large fraction
of the exposures will be directed towards the floor or to an upright image receptor,
and sometimes to other barriers (as in cross-table lateral image acquisitions). Barriers
that can intercept the unattenuated primary beam are considered primary barriers.
A protected control area for the technologist is required, with the ability to observe
and communicate with the patient. The viewing window of the control booth should
be of similar attenuation as the wall and large enough to allow unobstructed viewing. The configuration of the room should not depend on the control area shielding
as a primary barrier, and in no situation should there be an unprotected direct line
of sight from the patient or x-ray tube to the x-ray machine operator or to loaded
screen-film or CR cassettes, regardless of the distance from the radiation sources.
Also, the switch that energizes the x-ray tube should be installed so that the operator
cannot stand outside of the shielded area and activate the switch.
Fluoroscopic imaging systems are typically operated over a range of 60 to 120 kV.
Since the image receptor is designed to also be a primary barrier, only secondary
radiation barriers need to be considered in the shielding design. In some cases, a
radiographic and fluoroscopic combined unit is installed, and the shielding requirements are based on the combination of the workloads of both units. In this case, the
radiographic room issues discussed above must also be taken into account.
Interventional facilities include angiography and vascular interventional, neuroangiography, and cardiovascular and electrophysiology imaging suites. Like other fluoroscopic rooms, the walls, floors, and ceilings of interventional suites are considered
to be secondary radiation barriers. However, they may have multiple x-ray tubes and
the procedures often require long fluoroscopy times, and include cine and digital
fluorography image sequences that have large workload factors and so the rooms
may require more shielding than general fluoroscopy rooms.
A dedicated chest radiography installation has the x-ray tube directed at the image
receptor assembly on a particular barrier all of the time. Since the receptor can be
used at various heights above the floor, the area behind the image receptor from the
finished floor to a height of 2.1 m must be considered to be a primary barrier. All
other areas in this room are secondary barriers, and any portion of the wall that the
primary beam cannot be directed toward can also be considered a secondary barrier.
858
Mammography employs a very low kV in the range of 25 to 35 kV, and the breast
support provides the primary barrier for the incident radiation. Thus, radiation barriers
protect from secondary radiation only, and given the low x-ray energies and the small
volume of tissue irradiated, often all that is needed in a permanent mammography
room is the installation of a second layer of a gypsum wallboard. Doors for mammography rooms might need special consideration, because wood attenuates much less than
typical gypsum wallboard; a metal door may be advisable. Regarding operator safety,
mammography machines have lead acrylic barriers to protect the control area.
CT uses a collimated x-ray fan beam intercepted by the patient and by the detector array; thus, only secondary radiation reaches protective barriers. The x-ray tube
voltage used for most scans is 120 kV, over a range of 80 to 140 kV. As mentioned in
Chapter 10, modern wide-beam multirow detector CT scanners (MDCTs) make more
efficient use of the radiation produced by the x-ray tubes than did the now-obsolete single
detector row scanners. MDCTs can perform extensive scanning without exceeding
their heat limits, allowing more procedures per day, more contrast phases per procedure, and scans that cover more patient anatomy. Although the amount of scattered radiation per equivalent scan is not significantly more than that produced by a
single-row scanner, a large number of scans per day can require a greater thickness of
shielding for walls, and perhaps even additional shielding of floors and ceilings. Secondary scatter emanating from the scanner is not isotropic, as there are much higher
radiation levels along the axis of the patient table than in the direction of the gantry
(see Fig. 21-14). Assuming an isotropic scatter distribution is conservative in terms
of the amount of shielding for the various barriers in a room, it provides flexibility for
future CT scanner installations when different orientations are considered.
Mobile radiography and fluoroscopy systems (used in situations in which patients
cannot be transported to fixed imaging systems or used in operating rooms or intensive care units) present a challenge for protecting nearby individuals. For bedside
radiography, protection is chiefly accomplished by maintaining a distance from
the source of radiation and keeping the primary beam directed away from anyone
FIGURE 21-11 Cost of lead (adapted from Fig. 2-3, NCRP Report No.147). Lead is commercially sold by
nominal weight in pounds per square foot (lb/ft2). Equivalent thickness in inches and millimeters are also stated.
The height of each bar is the relative cost of lead sheet compared to 2 lb/ft2, normalized to a value of 1.
859
nearby. For mobile fluoroscopy, all individuals within 2 m of the patient should wear
protective aprons, and if available, protective mobile shielding should be positioned
between the patient and attending personnel.
A bone mineral densitometry x-ray system typically uses a well-collimated scanning
beam and, due to the low beam intensity and correspondingly low x-ray scatter, such
a scanner will not produce scattered radiation levels above 1 mGy per year at 1 m for
a busy facility, which is the shielding design goal for a fully occupied, uncontrolled
area (NCRP, 2004). Thus, structural shielding is not required in most situations. The
control console, however, should be placed as far away as practicable to minimize
exposure to the operator.
Dental and veterinary x-ray facilities require special consideration depending on
the scope and unique attributes of the procedures. The NCRP has published reports
that describe shielding and radiation protection requirements for these facilities. For
these and all other applications and future developments, sources of ionizing radiation should be evaluated by a qualified expert in order to determine the type and
nature of the shielding required in the facility.
Shielding Materials
X-ray shielding is accomplished by interposing an attenuating barrier between the
source(s) of radiation and the area to be protected in order to reduce the exposures
to below acceptable limits. The thickness of shielding needed to achieve the desired
attenuation depends on the shielding material selected. Lead is the most commonly
used material because of its high-attenuation properties and relatively low cost.
Commercially available thicknesses of lead sheet are commonly specified in nominal
weight per area (in pounds per square foot), with corresponding thicknesses specified
in inches and millimeters, as shown in Figure 21-11. The actual masses per area are
considerably less than the nominal values. The least thickness commonly installed is
2 lb/ft2 equal to 1/32 inch or 0.79 mm; there is little cost savings from using a lesser
thickness. Other thicknesses and relative costs are compared in the figure. For typical shielding installations, the lead sheet is glued to a sheet of gypsum wallboard and
installed with nails or screws on wood or metal studs. Where the edges of two lead
sheets meet, continuity of shielding must be ensured by overlapping lead, as well as
for gaps and inclusions in the wall (e.g., electrical junction boxes and switches.).
Other shielding materials are also used, such as gypsum wallboard, concrete,
glass, leaded glass, and leaded acrylic. Gypsum wallboard (sheetrock) is used for wall
construction in medical facilities, and a nominal 5/8 inch thickness (14 mm minimum) is most often used. While there is little protection provided at higher energies, significant attenuation occurs at the low x-ray energies used for mammography.
Because of possible non-uniformity of the gypsum sheets, it is prudent to specify
an extra layer (e.g., two sheets of wallboard) when using this material for shielding.
Concrete is a common construction material used in floors, wall panels, and roofs,
and is usually specified as standard-weight (147 lb/ft3, 2.4 g/cm3) or light-weight
(115 lb/ft3, 1.8 g/cm3). The concrete density must be known to determine the thickness needed to provide the necessary attenuation. When concrete is poured on a
ribbed-profile steel deck, the thickness is not constant and the minimum concrete
thickness should be used for attenuation specifications. Glass, leaded glass, and
leaded acrylic are transparent shielding materials. Ordinary plate glass may be used
when protection requirements are low; its attenuation may be increased by laminating two or more 6 mm glass sections. More common and useful are leaded glass
(glass with a high lead content) and leaded acrylic (impregnated with lead during
860
manufacturing) that are specified in various thicknesses of lead equivalence, such as

0.5, 0.8, 1.0, and 1.5 mm.
Computation of X-ray Imaging Shielding Requirements

Terminology
As stated above, the shielding design goals, P, are 0.1 mGy/wk for controlled areas
and 0.02 mGy/wk for uncontrolled areas. The distance from the radiation source to
the nearest approach to the barrier of the sensitive organs of a person in the occupied
area must be chosen; the point of closest approach to the barrier is assumed to be 0.3
m for a wall, 1.7 m above the floor below, and, for transmission through the ceiling,
at least 0.5 m above the floor of the room above, as shown in Figure 21-12.
The occupancy factor, T, for an area is defined as the average fraction of time
that the maximally exposed individual is present while the x-ray beam is on. The
maximally exposed individuals will usually be employees of the facility, or residents or employees of an adjacent facility. Recommended values for T are listed in
Table 21-3, for use when information about actual occupancy for a specific situation
is not known. The occupancy factor modifies the shielding design goal allowable at a
given point by 1/T; in other words, the attenuation of a barrier must lower the radiation to a level given by the ratio P/T.
The workload (W) is the time integral of the x-ray tube current in milliampereminutes over a period of 1 week (mA-min/wk). In NCRP Report No. 147,
a normalized average workload per patient, Wnorm, is also described; it includes
multiple exposures depending on the type of radiographic exam and clinical goal.
The total workload per week, Wtot, is the product of Wnorm and the average number
of patients per week (N).
Unlike earlier shielding methods that assumed a single, fixed high kV value for
the workload, the workload distribution described in NCRP Report No. 147 is a function of the kV, which is spread over a wide range of operating potentials for extremity examinations (e.g., about 1/3 of the total exams in a general radiographic room)
at 50 to 60 kV, abdominal exams at 70 to 80 kV, and chest exams at greater than
FIGURE 21-12 A. Floor plan of a general purpose radiographic room shows the location of the x-ray source,
orientation and location of the x-ray beam for various acquisitions, and the adjacent areas with typical occupancy factors, T; shielding design goals, P; and indication of primary and secondary barriers. Distances to each
barrier are determined for the primary beam, leakage radiation from the x-ray tube, and scattered radiation
from the patient. For secondary radiation, the closest distance is used. B. The elevation diagram provides
information to determine the distances to adjacent areas above and below the x-ray room to be shielded. The
floor-to-floor heights in multi-story buildings must be known.
861
TABLE 21-3 SUGGESTED OCCUPANCY FACTORS FOR ADJACENT AREAS

OCCUPANCY LEVEL
LOCATION
Full
Administrative or clerical offices; laboratories, pharmacies, and other work areas fully
occupied by an individual; receptionist areas,
attended waiting rooms, childrens indoor play
areas, adjacent x-ray rooms, film-reading areas,
nurses stations, x-ray control rooms
Partial
Rooms used for patient examinations and

treatments
Corridors, patient rooms, employee lounges,
staff rest rooms
Corridor doors
1/2
Public toilets, unattended vending areas, storage rooms, outdoor areas with seating, unattended waiting rooms, patient holding areas
1/20
Outdoor areas with only transient pedestrian

or vehicular traffic, unattended parking lots,
vehicular drop-off areas (unattended), attics,
stairways, unattended elevators, janitors closets
1/40
Occasional
OCCUPANCY FACTOR (T)
1/5
1/8
National Council on Radiation Protection and Measurements. Structural shielding design for medical x-ray
imaging facilities. NCRP Report No. 147. National Council on Radiation Protection and Measurements,
Bethesda, MD, 2004.
100 kV (with reduced tube current-time product). Workload distributions are specific for a given type of radiological installation, including radiographic room (all
barriers, chest bucky, floor, or other barriers), fluoroscopy tube (R&F room), radiographic tube (R&F room), chest room, mammography room, cardiac angiography,
and peripheral angiography. For shielding design, the magnitude of the workload is
less important than the distribution of the workload as a function of kV, since attenuation properties of the barriers have a strong kV dependence.
The use factor (U) is the fraction of the primary beam workload that is directed
toward a given primary barrier, and will depend on the type of radiographic room
and the orientation of the equipment. In a dedicated chest room, the primary barrier
has U = 1, since the x-ray beam is always directed toward the wall-mounted chest
receptor, and all other walls are secondary radiation barriers. In a general radiographic room, U must be estimated for the types of procedures and the amount of
time the x-ray beam will be used for a specific orientation. Most often, the x-ray
beam is pointed at the floor for acquiring images with patients on the table, occasionally pointed at a wall for a cross-table acquisition, and at a chest receptor stand for
a combined room. For a general radiographic room, a survey of clinical sites (NCRP
Report No. 147) yielded, for radiography not using a vertical chest receptor, U = 0.89
for the floor, U = 0.09; for the cross-table lateral wall, and U = 0.02 for an unspecified
wall; for procedures using a vertical chest receptor stand, U = 1 for the chest receptor
stand; and U = 0 for the ceiling and the control area barrier in that same room. For
fluoroscopy and mammography, the primary beam stop is the image receptor assembly, so U = 0 and only secondary radiation must be considered.
The primary barrier, found in radiographic, dedicated chest, and radiographicfluoroscopic rooms, is designed to attenuate the primary beam to the shielding
design goal. Unshielded primary air kerma at the point to be protected per week is
dependent on the average number of patients per week (N), the use factor (U) for
862
that barrier, the primary air kerma per patient at 1 m, and the distance to the point
(inverse square law correction is applied).
The secondary barrier is designed to attenuate the unshielded secondary air kerma
from leakage and scatter radiation to the shielding design goal (or less). All walls not
considered primary barriers are secondary barriers. Scattered radiation from the patient
increases with the intensity and area of the useful beam, and is a function of the scattering angle. The total contribution from unshielded secondary air kerma, proportional
to Wtot, is calculated from the clinical workload distributions, determined at 1 m, and
modified by the inverse square law to the distance of the point to be protected, similar
to that described for primary radiation. In certain orientations such as lateral acquisitions, the distance from the scattering source (the patient) will be different than the distance from the leakage radiation source (the x-ray tube) to the point to be protected.
Example Shielding Calculation

Room diagrams of a radiographic room are shown in Figure 21-12, with distances
in meters from each of the radiation sources (primary, x-ray tube leakage, and scatter from the patient), as well as adjacent areas that are defined in terms of T, P, and
primary or secondary barriers.
There are many complexities involved in calculating the necessary attenuation of the
primary and secondary barriers to achieve the shielding design goals that are beyond
the scope of this book. Many of the details regarding workloads, workload distributions,
normalized workloads per patient (Wnorm), field size area, attenuation of primary radiation by the image receptor, and other nuances are incorporated into shielding thickness
charts found in NCRP Report No. 147, which display the required shielding thicknesses
for the various barriers as a function of N 3 T/(P 3 d2), where as stated previously, N
is the number of patients per week, T is the occupancy factor (see Table 21-3), P is the
shielding design goal in mGy/wk, and d is the distance in meters from the radiation
source to the point to be protected. In abbreviated notation, this is NT/Pd2. These charts,
which assume specific normalized workloads per patient (Wnorm) as described in NCRP
Report No. 147, are provided for radiographic and radiographic/fluoroscopic rooms for
both lead and concrete shielding materials. If the site-normalized workload per patient
(Wsite) is different, then the values of NT/Pd2 are scaled by Wsite/Wnorm. Primary beam use
factors U are identified by specific curves in each chart for primary barriers.
Three of these charts are shown in Figure 21-13 for a radiographic room, considering (A) no primary beam preshielding, (B) primary beam preshielding, and
(C) secondary radiation. The required lead thickness in millimeters is a function of
NT/Pd2. The term preshielding refers to attenuation of the primary beam by the image
receptor. The graphs for primary beam preshielding consider the attenuation of the
image receptor in the radiographic table or wall-mounted cassette holder due to the grid,
cassette, and support structures, equivalent to 0.85 mm lead, and for cross-table laterals
attenuation due to the grid and cassette, equivalent to 0.3 mm lead. Note the decreased
requirement for lead thickness with primary beam preshielding (e.g., for the chest bucky
wall barrier and NT/Pd2 = 1,000, without preshielding the thickness is 1.95 mm lead
and with preshielding is 1.13 mm lead, as shown in Fig. 21-13A,B, respectively).
For determination of barrier thicknesses using NT/Pd2 methods, scale drawings
of the room, including elevation drawings, location of the equipment, designations
of adjacent areas as controlled or uncontrolled, occupancy factors of adjacent areas,
and distances to adjacent areas are necessary, in addition to the average number of
patients, N, per week. Charts are selected according to room type and shielding materials. For example, using the diagrams in Figure 21-12, if N = 120 patients per week
for a general radiographic room, then NT/Pd2 can be calculated for each barrier and
863
FIGURE 21-13 Required lead barrier

thickness for a radiographic room as a
function of NT/Pd2, using Wnorm = 0.6 mA
min patient21 for the chest bucky and
1.9 mA min patient1 for the floor and
other barriers. A. No preshielding for primary barriers. B. Preshielding for primary
barriers. C. Secondary lead barriers. In each
of the graphs, the horizontal dotted lines
indicate the common lead thicknesses of
0.79 and 1.58 mm, corresponding to
2 lb/ft2 and 4 lb/ft2 lead sheet. (Adapted
from Fig 4.5AC, NCRP Report No. 147.)
864
the required lead thickness is determined from the graph. For the chest bucky wall
primary radiation barrier with T = 0.2, P = 0.02 mGy/wk, and d = 2.4 m, NT/Pd2 = 120
3 0.2/(0.02 3 2.42) = 208 and, using the blue curve labeled chest bucky wall, the lead
thickness required is about 1.3 mm to achieve the shielding design goal. Therefore,
the primary barrier to be installed is specified as 4 lb/ft2 (1.6 mm) over the indicated
area, which is from the finished floor to a height of 2.1 m (7 ft), with lateral margins
around the image receptor of at least 0.5 m (e.g., a leaded drywall sheet 4 ft wide).
Shielding for secondary barriers is determined in a similar manner; however,
because the total amount of secondary radiation emanates from a variety of locations and distances, the calculations can be very complex. A conservative approach
is to assume that all of the secondary radiation in the room emanates from the closest distance to the barrier, which would increase the shielding thickness. Thus,
for the wall adjacent to the chest wall bucky designated as a secondary barrier,
the closest distance is 1.5 m, and NT/Pd2 = 120 3 0.2/(0.02 3 1.52) = 533. From
Figure 21-11C, the thickness of lead (using the chest bucky secondary wall curve)
is about 0.6 mm, and therefore the secondary barrier is specified as 2 lb/ft2, installed
continuously and seamlessly with the adjacent primary radiation barrier. For the
control area barrier with T = 1, P = 0.1 mGy/wk, d = 2.5 m, NT/Pd2 = 120 3 1/
(0.1 3 2.52) = 192, and using the curve labeled secondary wall, the lead thickness
is 0.15 mm, so 2 lb/ft2 of lead is specified. For the control area window, leaded acrylic
is often used, and would be specified as having an equivalent attenuation as the surrounding wall of 2 lb/ft2 or 0.8 mm lead. Note that in these situations, the conservative
approach ensures more than adequate protection for individuals in adjacent areas, and
allows for increased room workload in the future. The shielding for each other radiation
barrier, including doors, the floor, and ceiling, is similarly calculated and specified.
In walls, the shielding (e.g., leaded drywall, leaded doors, leaded acrylic, and
sheet lead) is installed from the finished floor to a height of 2.1 m (7 ft) with overlapping lead at the seams of at least 1 cm, and lead backing at electrical outlets
and other access points (electrical junction boxes, plumbing, etc.) into the wall to
ensure continuous protection. Attachment screws used for mounting leaded drywall do not need lead backing (the steel that displaces the lead combined with
the length of the screw provides adequate attenuation), but care must be taken to
ensure that the screws that penetrate the lead sheet remain in place to maintain
shielding integrity.
Surveys after Shielding Installation

After installation of shielding, a radiation protection survey should be performed
by a qualified expert to ensure the adequacy of the installed shielding. The survey
should verify that the barriers are contiguous and free of voids and defects and provide sufficient attenuation to meet the relevant shielding design goal divided by the
occupancy factor (P/T). Visual inspection during installation should be part of the
evaluation, followed by transmission measurements using a gamma-ray source of
suitable energy (e.g., Tc-99m) and a detector with high sensitivity and fast response
(e.g., GM survey meter). Alternatively, the x-ray source in the room can be used to
generate a primary beam and secondary radiation, and measurements performed
with a suitable high-sensitivity integrating survey meter.
Computed Tomography Scanner Shielding

For a CT scanner, all walls in the room are secondary barriers, because the detector
array within the gantry is the primary radiation barrier. The scatter distribution emanating from the scanner is highly directional, being highest along the scanner axis
865
(Fig. 21-14). Because most CT scans are acquired at 120 and 140 kV with a highly
filtered beam, the scattered radiation is highly penetrating. There are three methods
that can be used to determine the shielding requirements, one based upon CTDIvol,
another based upon typical dose length product (DLP) values per acquisition, and
the third based on the measured scatter distribution isodose maps provided by the
manufacturer. For instance, using typical values for DLP, the scatter intensity (air
kerma in mGy/acquisition) at 1 m is estimated for head and body scans, and then
multiplied by the number of acquisitions (much greater than the number of patients
due to multiple acquisitions being performed of many patients) expected over a
typical work week for an 8-h shift (the busiest shift if the scanner is operated over
extended hours). This is the total unshielded peak air kerma at 1 m from the gantry.
For each barrier, the distance from the gantry to the calculation point beyond the
barrier is determined, and the inverse square law is applied. A thickness of shielding is then determined to reduce the radiation level to the design goal modified by
the inverse of the occupancy factor (P/T) by the use of specific transmission charts
(Fig. 21-15).
FIGURE 21-14 Secondary (scatter
and leakage) radiation distributions
for a 32-cm-diameter PMMA phantom, 40-mm collimation, 64-channel
MDCT, 140 kV, 100 mA, 1 s scan.
Actual secondary radiation levels must
be scaled to the techniques used for
the acquisition. A. Horizontal isodose
scatter distribution at the height of
the isocenter. B. Elevational isodose
scatter distribution; the distribution is
vertically symmetric above and below
the level of the table.
866
Example Barrier Calculation for a CT Scanner

A CT scanner performs 125 head procedures/week and 150 body procedures/week
using a tube voltage of 120 kV, and produces a scatter air kerma of 63.8 mGy/wk
at 1 m from the gantry isocenter. A barrier that is 3.3 m (11 ft) from the gantry isocenter to an uncontrolled corridor with P = 0.02 mGy/wk and an occupancy factor
T = 0.2 requires a transmission factor to reduce the radiation to a level less than
P/T = 0.02/0.2 = 0.1 mGy/wk. At 3.3 m, the unshielded air kerma is 63.8/3.32 =
5.86 mGy/wk, and the transmission factor through the lead must therefore be less
than 0.1/5.86 = 0.017 Consulting Figure 21-15, this transmission factor requires
a minimum of 1.0 mm lead. This is greater than 2 lb/ft2, so 4 lb/ft2 is specified for
this barrier. The shielding for other barriers of the CT scanner room is calculated
similarly. For the floors and ceilings, a transmission chart with varying thicknesses
of concrete should be used; if the thickness of concrete is insufficient, a sheet of
lead should be installed above or below the concrete.
For multirow detector array scanners with collimator widths of 40 mm (and in
some cases greater, up to 160 mm), the amount of scatter per gantry rotation for a
specific kV and mAs will typically be much higher than that of a single-slice scanner.
However, because of the increased axial coverage per rotation, the total mAs per scan
will be a fraction of the single-slice scanners, on the order of two to four times lower,
and thus, the amount of scatter per scan will be comparable. On the other hand, the
acquisition speed allows a larger number of patients to be scanned, which can significantly increase the total amount of scatter that must be considered for shielding
calculations. It is therefore prudent to determine the shielding for the peak patient
workloads expected over a given work shift (8 h is typical, but in some cases this can
be up to 12 h on the job) to ensure adequate protection from secondary radiation in
adjacent areas. Dose-reducing features on CT scanners, such as automatic tube current modulation, may reduce shielding requirements in the future.
FIGURE 21-15 Lead thickness as a function of transmission factor for secondary radiation from a CT scanner operated at 120 kV. (National Council on Radiation Protection and Measurements. Structural shielding
design for medical x-ray imaging facilities. NCRP Report No. 147. National Council on Radiation Protection and
Measurements, Bethesda, MD, 2004.)
867
PET/CT and SPECT/CT Shielding

Most PET today is performed using the radiopharmaceutical F-18 fluorodeoxyglucose (FDG). In PET imaging with F-18 FDG, the radiopharmaceutical is administered to a patient, the patient rests for a period of about 45 to 90 min to allow uptake
by cells and washout from other physiologic compartments, the patient urinates to
remove activity from the bladder, and then is imaged. The installation of shielding
must be considered for the room containing the PET/CT system and for the uptake
rooms where patients rest after radiopharmaceutical administration and before imaging. It is common for a PET/CT facility to have three uptake rooms per PET/CT
system because a modern PET/CT system can scan about three patients an hour.
Required shielding depends upon the patient workload, the activity administered per
patient, the uses of areas near the PET/CT room and uptake rooms, and the distances
to these nearby occupied areas. Because of the high energies (511 keV) of annihilation photons, the shielding thicknesses can be very large and consideration must be
given to whether the building can support the weight of the shielding. The amount
of shielding needed can be minimized by placing the PET/CT and uptake rooms on
the lowest floor (no shielding needed below) or on the top floor (no shielding needed
above) and against an exterior wall or walls, thereby reducing or eliminating the need
for shielding of one or two walls. Placing the uptake rooms adjacent to each other
also reduces the weight of shielding needed. The amount of needed shielding can be
further reduced by designing the area so that areas of low occupancy (e.g., storage
rooms, mechanical and electrical equipment rooms, and bathrooms) are immediately
adjacent to the uptake and PET/CT rooms. The shielding design methods for PET/CT
facilities are described in AAPM Report No. 108, PET and PET/CT Shielding Requirements (AAPM, 2005).
The shielding design for a PET/CT or a SPECT/CT facility must also consider the
radiation from the x-ray CT system. In most PET/CT installations, the thick shielding
for the annihilation photons is more than sufficient for the lower-energy secondary
radiation from the x-ray CT system. SPECT imaging rooms are often not shielded;
for a SPECT/CT, a shielding calculation should be performed for the x-ray CT system
and shielding should be installed if needed. In some SPECT/CT scanners, a conebeam detector is utilized to acquire the CT projection data, and will have scatter
distributions that might require specific considerations when determining the barrier thickness. If a nuclear medicine imaging room is near a PET/CT room or a PET
uptake room, shielding of greater thickness than needed for the protection of people
may be required so that the high-energy photons emanating from patients do not
interfere with imaging procedures in the nearby room.
21.6 Radiation Protection in Diagnostic and Interventional
X-ray Imaging
Personnel Protection in Medical X-ray Imaging

During CT, radiographic, and remote fluoroscopic procedures, medical staff are seldom
in the room with the patient while x-rays are being produced, although there may be
exceptions for pediatric and critically ill patients and those who are incapable of following instructions. During most procedures, staff are protected from radiation by the structural shielding installed in the walls of the room. However, physicians and assisting staff
are in the room for most procedures utilizing fluoroscopy, particularly angiographic and
interventional procedures, and during some CT-guided diagnostic and interventional
868
procedures. Figure 21-16 shows the varying intensity of stray radiation, primarily scattered x-rays from the patient, during fluoroscopy. Understanding the spatial pattern of
stray radiation can help staff to reduce the doses by where they stand during the procedure and how they position movable shielding. In particular, when the x-ray beam is in
a lateral angulation, the stray radiation intensity is much higher on the side of the patient
toward the x-ray source than on the side toward the image receptor. The x-ray doses to
the heads and arms of staff tend to be least when the x-ray tube is beneath the patient.
Several items provide protection for staff during a fluoroscopic or radiographic
imaging procedure. The first and foremost is the protective apron worn by all individuals who must work in the room when the x-ray tube is operated. Lead equivalent thicknesses from 0.25 to 0.50 mm are typical. Usually, the lead is in the form of
a rubber material to provide flexibility and handling ease. Aprons protect the torso
of the body and the upper legs and are available in designs with only frontal shielding
or with wrap-around shielding, the latter being important when the back is exposed to
the scattered radiation for a considerable portion of the time. Greater than 90% of the
scattered radiation incident on the apron is attenuated by the 0.25-mm thickness at
standard x-ray energies (e.g., tube voltage less than 100 kV). Aprons of 0.35 and 0.50
mm lead equivalents give greater protection (up to 95% to 99% attenuation), but weigh
50% to 100% more than the 0.25-mm lead equivalent aprons. Since the k-absorption
edge of lead is at 88 keV and the average energy of typical diagnostic x-rays is about
35 to 40 keV, materials in the middle range of atomic numbers (Z) such as tin
and barium are more efficient per unit weight in absorbing those x-rays than is
lead. Several types of non-lead and lead composite aprons for shielding personnel
from diagnostic x-rays are available (Murphy et al., 1993; Zuguchi et al., 2008).
The shielding capabilities of the lighter aprons are commonly specified in mm
lead equivalent, but such a specification is valid only for a specific x-ray energy
spectrum. These lighter aprons are typically adequate for x-rays generated at tube
potentials up to 120 kV, but their effectiveness compared to conventional lead
aprons declines at tube potentials above about 100 kV. For long fluoroscopic procedures, the weight of the apron may become a limiting factor in the ability of the
physician and the attending staff to complete the case without substitutions. Some
FIGURE 21-16 Dose rates from scattered radiation during fluoroscopy, with the x-ray beam in PA (A) and lateral (B) orientations. These diagrams show the decrease in dose rate with distance from the location where the
x-ray beam enters the patient. When the x-ray tube is beneath the patient (A), the highest scatter intensity is to
the lower part of the operators body. When the x-ray beam has a lateral angulation (B), the scatter intensity is
much less on the side of the patient toward the image receptor. (From: National Council on Radiation Protection
and Measurements. Radiation dose management for fluoroscopically guided interventional medical procedures.
NCRP Report No. 168. National Council on Radiation Protection and Measurements, Bethesda, MD, 2010a.)
869
apron designs, such as skirt-vest combinations, place much of the weight on the
hips instead of the shoulders. The areas not covered by the apron include the arms,
lower legs, the head and neck, and the back (except for wrap-around aprons).
Aprons also do not protect the thyroid gland or the eyes. Accordingly, there are
thyroid shields and leaded glasses that can be worn by the personnel in the room. The
leaded thyroid shield wraps around the neck to provide attenuation similar to that of
a lead apron. Leaded glasses attenuate the incident x-rays to a lesser extent, typically
30% to 70%, depending on the content (weight) of the lead. Unfortunately, their
weight is a major drawback. Normal, everyday glasses provide only limited protection, typically much less than 20% attenuation. During fluoroscopy, the operator
commonly looks at the display monitor while x-rays are produced and the x-rays
typically strike his or her head from the side and below. Therefore, the protective
glasses or goggles should provide shielding on the sides, or the glasses should be of
wrap-around designs. Whenever the hands must be near the primary beam, protective gloves of 0.5-mm thick lead (or greater) should be considered when use does not
interfere with the dexterity required to carry out the procedure.
In high workload angiographic and interventional laboratories, ceiling-mounted,
table-mounted, and mobile radiation barriers are often used (Fig. 21-17). These devices
are placed between the location where the x-ray beam intercepts the patient and the
personnel in the room. The ceiling-mounted system is counterbalanced and easily
positioned. Leaded glass or leaded acrylic shields are transparent and usually provide
greater attenuation than the lead apron.
Only people whose presence is necessary should be in the imaging room during
imaging. All such people must be protected with lead aprons or portable shields. In
addition, no persons, especially individuals who are pregnant or under the age of 18
years, should routinely hold patients during examinations. Mechanical supporting
or restraining devices must be available and used whenever possible. In no instance
should the holders body be in the useful beam, and it should be as far away from the
primary beam as possible.
Protection of the Patient in Medical X-ray Imaging

The two main methods for limiting the radiation doses to patients in medical imaging
are to avoid unnecessary examinations and to ensure the doses from examinations
are no larger than necessary. The goal of a radiological examination should be to
produce images of adequate quality for the clinical task, not images of unnecessarily
high quality if that increases the radiation dose to the patient. On the other hand,
using too low a dose can also be harmful, if it results in images that cause a clinical
error or must be repeated.
Tube Voltage and Beam Filtration

An important goal in diagnostic imaging is to achieve an optimal balance between
image quality and dose to the patient. Increasing the kV will result in a greater transmission (and therefore less absorption) of x-rays through the patient. Even though the
air kerma (or exposure) per mAs increases as the kV is increased, an accompanying
reduction in the mAs to produce a similar signal to the image receptor will decrease
the incident exposure to the patient. Unfortunately, there is a concomitant reduction
in image contrast due to the higher effective energy of the x-ray beam. Within limits,
this compromise is acceptable. Therefore, the patient exposure can be reduced by
using a higher kV and lower mAs. However, in procedures such as angiography and
CT angiography in which contrast material is to be visualized and not soft tissue,
870
FIGURE 21-17 Shielding used

with a C-arm fluoroscope. The
optimal position of the movable
ceiling-mounted transparent shield
is approximately perpendicular to
and in contact with the patient. A
table-mounted shield (seen in front
and to the left of the operator) protects the lower part of his body. The
operator is looking at the display
monitors and scattered radiation
striking his face is from the side and
below. (Source: National Council on
Radiation Protection and Measurements. Radiation dose management
for fluoroscopically-guided interventional medical procedures. NCRP
Report No. 168. National Council on
Radiation Protection and Measurements, Bethesda, MD, 2010a.)
greater beam filtration and lower kV can dramatically improve the contrast to noise
ratio, which can result in a lower dose to the patient for these applications.
Filtration of the polychromatic x-ray energy spectrum can significantly reduce
exposure by selectively attenuating the low-energy x-rays in the beam that would otherwise be absorbed in the patient with little or no contribution to image formation.
These low-energy x-rays mainly impart dose to the skin and shallow tissues where the
beam enters the patient. As the beam filtration is increased, the beam becomes hardened (the average photon energy increases) and the dose to the patient decreases
because fewer low-energy photons are in the incident beam. The amount of filtration
that can be added is limited by the increased x-ray tube loading necessary to offset the
reduction in tube output, and the decreased contrast that occurs with excessive beam
hardening. Except for mammography, a moderate amount of beam filtration does not
significantly decrease subject contrast. The quality (effective energy) of the x-ray beam
is assessed by measuring the HVL. State and federal regulations require the HVL of
the beam to exceed a minimum value (see Chapter 6, Table 6-3, and for mammography systems, Chapter 8, Table 8-2). Adequate filtration helps to avoid or minimize
skin injuries from prolonged fluoroscopically guided interventional procedures.
Field Area, Organ Shielding, and Geometry

Restriction of the field size by collimation to just the necessary volume of interest is
an important dose reduction technique. While it does not significantly reduce the
entrance dose to the area in the primary beam, it reduces the volume of tissue in
the primary beam and thus the energy imparted to the patient. It also reduces the
amount of scatter and thus the radiation doses to adjacent organs. From an image
quality perspective, the scattered radiation incident on the detector is also reduced,
thereby improving image contrast and the signal-to-noise ratio.
When possible, particularly radiosensitive organs of patients undergoing radiographic examinations should be shielded. For instance, when imaging a limb (such
as a hand), a lap apron should be provided to the patient. When the gonads are
871
in the primary beam, gonadal shielding should be used to protect the gonads when
the shadow of the shield does not interfere with the anatomy under investigation.
Because the gonadal shield must attenuate primary radiation, its thickness must be
equivalent to at least 0.5 mm of lead. In any situation, the use of patient protection must not interfere with the examination. Other organs that may benefit from
shielding if they must be in the primary beam include the lenses of the eyes and
breast tissue in female children and young women. Care must be taken regarding
proper placement of organ-specific shielding; distracted technologists have been
known to place it on the x-ray receptor side of the patient instead of the x-ray
tube side. Care also must be taken to ensure that gonadal and other organ-specific
shielding does not block an active sensor for the automatic exposure control system. When critical organs must be in the beam, their doses can be reduced by
orienting the beam so that the critical organ is on the side of the patient away from
the x-ray tube.
Increasing the distance from the x-ray source to the patient (source-to-object
distance [SOD]) helps reduce dose. As this distance is increased, a reduced beam
divergence limits the volume of the patient being irradiated, thereby reducing the
integral dose. Increasing this distance also reduces entrance dose due to the inverse
square law. The exposure due to tube leakage is also reduced since the distance from
the tube to the patient is increased, although this is only a minor consideration.
With C-arm fluoroscopic systems that have fixed source-to-image receptor distances,
patient dose is reduced by increasing the SOD as much as possible and moving
the patient as close to the image receptor as possible. Federal regulations (21 CFR
1020.32[g]) specify a minimum patient (or tabletop) to focal spot distance of 20 cm.
Furthermore, maintaining at least 30 cm is strongly recommended to prevent excessive radiation exposure.
X-ray Image Receptors

For film-screen image receptors, the speed of the image receptor determines the
number of x-ray photons and thus the patient dose necessary to achieve an appropriate film darkening. Relative speed values are based on a par speed screen-film
system, assigned a speed of 100; a higher speed system requires less exposure to
produce the same optical density (e.g., a 400-speed screen-film receptor requires
four times less exposure than a 100-speed system). Specialized film processing can
also affect the speed of the system. Either a faster screen or a faster film will reduce
the incident exposure to the patient. In the case of a faster film, the disadvantage
is increased quantum mottle and, in the case of a thicker intensifying screen, the
disadvantage is reduced spatial resolution. Thus, increasing the speed of the screenfilm image receptor reduces patient exposure, but the exposure reduction possible is
limited by image-quality requirements. In modern film-screen imaging, a 400-speed
system is typically used for head and body imaging and a 100-speed system may be
used for imaging the extremities.
Digital radiography devices using photostimulable storage phosphor (PSP) plates,
phosphor plates optically coupled to CCD cameras, or flat-panel thin-film transistor
array image receptors have wide exposure latitudes. Postacquisition image scaling
(adjustments to brightness and contrast) and processing modify the image for optimal
viewing. Thus, these detectors can compensate (to a large degree) for under- and overexposure and can reduce retakes caused by inappropriate radiographic techniques.
However, underexposed images may contain excessive quantum mottle, hindering
the ability to discern contrast differences, and overexposed images result in needless
exposure to the patient. In comparison to screen-film receptors, PSP receptors are
872
roughly equivalent to a 200-speed screen-film receptor in terms of quantum mottle

and noise (e.g., for adult examinations of the abdomen and chest [Compagnone et al.,
2006; Seibert et al., 1994]), whereas digital flat-panel detectors have higher detection
efficiencies, equivalent to 400- to 600-speed screen-film receptors. Techniques for
extremity imaging with digital radiographic receptors should deliver higher exposure
levels (similar to extremity screen-film cassettes with speeds of 75 to 100). In pediatric imaging, there are many examinations and corresponding technique protocols
that allow low-dose, high-speed imaging; however, there are also situations requiring
high spatial resolution and high signal-to-noise ratio which, in turn, require higher
dose. The flexibility of the digital detector response can be used to advantage in these
situations to achieve optimal image quality at appropriate patient doses.
Conventional screen-film radiographic image receptors have an inherent safety
feature. If the film processor is operating properly and the kV is adequate, using an
excessive quantity of radiation to produce an image results in an overly dark film. On
the other hand, when digital image receptors are used for radiography, fluoroscopy,
and CT, excessive incident radiation may not have an adverse effect on image quality,
and unnecessarily high exposures to patients may go unnoticed. Routine monitoring of the exposures to the digital image receptors in radiography, particularly in
mobile radiography in which automatic exposure control is commonly not possible,
is therefore necessary as a quality assurance measure. Similarly, careful calibration of
the automatic exposure control system in radiography, calibration of the automatic
exposure rate control system in fluoroscopy, and attention to several factors, discussed below, in CT are necessary to avoid unnecessarily high doses to patients.
The image intensifier is a detector system with a wide dynamic range, whereby
the necessary entrance exposure is determined by a light-limiting aperture or the
electronic gain of a subsequent detector (e.g., TV camera). Entrance exposures to
the image intensifier as low as one R per image for fluoroscopy or as high as 4 mR
per image for digital subtraction angiography are possible. The amount of exposure
reduction is limited by the quantum statistics (i.e., too little exposure will produce
an image with excessive quantum mottle which hinders the ability to discern subtle,
low-contrast anatomical features).
Fluoroscopy
Fluoroscopy imparts a moderate fraction of the cumulative effective dose delivered
in medical imaging (~18% of the total in 2006), but imparts some of the largest tissue doses to individual patients. Even though the fluoroscopic techniques are quite
modest (e.g., 1 to 3 mA continuous tube current at 75 to 85 kV or higher mA, short
exposure pulsed fluoroscopy at 7.5 to 15 frames per second), a fluoroscopic examination can require several minutes and, in some difficult cases, can exceed hours of
on time. For example, a procedure using 2 mA at 80 kV for 10 min of fluoro time
(1,200 mAs) delivers an air kerma of 60 mGy at 1 m (assuming 50 Gy/mAs for 80
kV at 1 m). If the focal spot is 30 cm from the skin, the patient entrance air kerma is
(100/30)2 3 60 mGy < 670 mGy! In the United States, fluoroscopic patient entrance
air kerma (exposure) rates at 30 cm from the x-ray source for C-arm systems and at
1 cm above the table for systems with the x-ray tube fixed under the table are limited
to a maximum of 87.3 mGy/min (10 R/min) for systems with automatic brightness
control (ABC) and to 44 mGy/min (5 R/min) for systems without ABC. Some systems
have a high air-kerma-rate fluoroscopic (turbo) mode delivering up to 175 mGy/
min (20 R/min) output at this reference point, a factor of five to ten times greater than
the typical fluoro levels. This capability, which requires continuous manual activation of a switch by the operator (and produces a continuous audible signal), must be
used very judiciously. Cineangiography studies deliver high dose rates of 175 to 600
873
mGy/min (20 to 70 R/min) at high frame rates (15 to 30 frames per s) but typically
with short acquisition times. The patient entrance air-kerma rate depends on the
angulation of the beam through the patient, with the air-kerma rate increasing with
the thickness of the body traversed by the beam (e.g., the air kerma rate is greater for
lateral projections than for anteroposterior [AP] or posteroanterior [PA] projections).
Severe x-rayinduced skin injuries to patients have been reported following
prolonged fluoroscopically guided interventional procedures, such as percutaneous transluminal coronary angioplasty, radiofrequency cardiac catheter ablation,
vascular embolization, transjugular interhepatic portosystemic shunt placement,
and other percutaneous endovascular reconstruction (Koenig et al., 2001). The
cumulative skin doses in difficult cases or from a series of procedures of a single
area of the body can exceed 10 Gy, which can lead to severe radiation-induced
skin injuries (see Specific Organ System ResponsesSkin in Chapter 20).
Figure20-18 shows severe skin injuries following several fluoroscopically-guided
interventional procedures. Such injuries may not become fully apparent until many
weeks or even months after the procedure. It is recommended that each medical
facility determine the procedures that may impart doses that exceed the threshold
for skin injury. Prior to each such procedure, it is recommended that each patient
be screened for factors that may decrease the dose threshold for skin injury; these
include large radiation exposures to the same part of the body from fluoroscopically
guided interventional procedures or radiation therapy; some drugs; some hereditary
diseases affecting DNA repair; and possibly other diseases such as connective tissue disorders, diabetes mellitus, and hyperthyroidism (Balter et al., 2010; Koenig
et al., 2001). While large radiation doses may be unavoidable in difficult cases,
care should be taken during all examinations to employ techniques that will minimize the radiation exposure. Methods for minimizing the radiation exposure to the
patient were described in Chapter 9 (see the section entitled Dose Reduction).
Chapter 9 discusses the metrics of patient dose that are displayed by fluoroscopes.
Fluoroscopes manufactured since June 10, 2006, display the cumulative air kerma
at a reference point (for C-arm fluoroscopes, this reference point is 15 cm from the
isocenter of rotation, toward the x-ray source), typically the kerma-area product
or dose-area product, and the cumulative fluoro time. Older systems may display
only the cumulative fluoro time. Studies have shown that cumulative air kerma and
kerma area product correlate better with peak skin dose than does the total fluoro
time (Fletcher et al., 2002; Miller et al., 2003). Ideally, a map of skin doses would be
generated after each procedure, but this is not available at this time. Implementation
of this feature in future systems will greatly assist in dose management. Gafchromic films that self-develop can be placed between the patient and support table to
provide a postexamination map of surface peak skin dose. Dosimeters may be placed
on the patients skin to measure skin dose. In any dose-monitoring system, the realtime display of accumulated patient dose provides the physician with the ability to
modify the procedure, if necessary, to minimize the potential for radiation-induced
skin injuries. The dose metrics (e.g., cumulative air kerma, kerma-area product, or
total fluoro time) recorded at the completion of the procedures allow physicians to
identify patients who may be at risk for radiation-induced skin damage. After all procedures, areas of the patients skin that received large doses should be documented
in the patients medical records. The patients should be notified and arrangements
made for clinical follow-up if there is the potential for significant skin injury.
Physicians using or directing the use of fluoroscopic systems should be trained
and credentialed to ensure they understand the technology of fluoroscopy machines,
particularly the dose implications of the different modes of operation; the biological effects of x-rays; and the techniques used to minimize patient and personnel
874
exposure. In addition, the physicians should be familiar with radiation control regulations. NCRP Report No. 168 is an excellent resource for all radiation protection
aspects of fluoroscopically guided interventional procedures, including the management of radiation exposure to patient and staff (NCRP, 2010a).
Computed Tomography
As mentioned earlier in this chapter, the average annual effective dose to the population of the United States from medical imaging increased by a factor more than
five from about 1980 until 2006 and about half of this effective dose in 2006 was
from CT imaging. In 2001, it was reported in the medical literature that at least
some medical facilities in the United States were using adult technique factors for the
CT examinations of children and even infants (Patterson et al., 2001); it has been
estimated that this practice causes a risk of cancer to the patients on the order of ten
times the risk to adult patients (Brenner et al., 2001). In recent years, professional
societies, advisory bodies such as the ICRP and NCRP, and governmental agencies
have urged that efforts be made to reduce the doses to patients from medical imaging. Because of the large fraction from CT imaging and the relatively large dose per
examination, much of this effort should be focused on CT imaging.
In CT, many operator-selectable parameters determine the patient dose. These
include the kV and mA applied to the x-ray tube, time per x-ray tube rotation, detectortable motion pitch, and beam filter. As mentioned in Chapter 10, these are not usually
individually selected by a technologist for each patient procedure, but are instead specified in organ- and procedure-specific protocols, such as chest (routine), three-phase
liver, and brain (CTA). A technologist selects the appropriate protocol for a patient and
indication, although the technologist may have to modify the technique in the protocol
for patient-specific circumstances. Modern CT scanners provide the capability to protect
protocols with passwords, which should be restricted to only a few individuals. The use
of passwords can protect protocols from inadvertent or deliberate modification.
Most radiographic images, other than those acquired with mobile machines, are
acquired utilizing automatic exposure control (phototiming). However, older CT
scanners use a fixed kV and mAs selected by the technologist, regardless of the size
of the patient, unless the CT technologist changes them. The fixed techniques in CT
for a given protocol often lead to unnecessarily high mAs values for thinner patients,
particularly neonatal and young children. A way to avoid this unnecessary radiation
dose to the patient is to train the CT technologists to reduce the mAs (and perhaps
the kV) for thinner patients. A technique chart can be devised for each such CT scanner showing kV and mAs values for different diameter patients as a guide to achieving
good image quality at proper dose levels. Table 21-4 shows the dose reductions that
can be achieved if the CT protocols (mAs values) are adjusted to the patients size.
New scanners have the ability to modulate the tube current as a function of tube
rotation angle and axial position on the patient, to compensate for the attenuation
variations along the beam path through the patient. Some CTs use the localizer images
(scout scan or topogram) to determine the mA, whereas others also measure the
amount of transmitted radiation during the scan itself. Automatic tube current modulation can markedly reduce exposure without decreasing image quality. However,
each implementation of automatic tube current modulation requires information to
be provided to the CT regarding the amount of random noise in the resultant images
tolerated by the interpreting physicians. The various CT manufacturers specify this
differently (e.g., for one manufacturer, a parameter called noise index and the
reconstructed slice thickness must be selected, whereas for another manufacturer, a
parameter called the reference mAs and a setting of weak, average, or strong
875
TABLE 21-4 DOSE REDUCTION POTENTIAL IN CT

PATIENT DIAMETER (cm)
PERCENT mAs
mAs
14
0.3
16
0.6
18
1.2
20
2.2
22
4.2
13
24
7.9
24
26
15
45
28
28
85
30
53
160
32
100
300
34
188
564
36
352
1,058
38
661
1,983
40
1,237
3,712
Note: If a CT technique of 120 kV and 300 mAs delivers good image quality (low
noise images) for a 32-cm-diameter patient, the same image quality (signal to
noise ratio [SNR]) can be achieved by reducing the mAs to the levels indicated
for patients smaller than 32, and increasing the mAs for patients larger than 32
cm. The large mA values for very large diameters are not possible; alternatively
a higher kV could be used. Large patients have greater amounts of abdominal
fat that results in improved conspicuity of organ boundaries, and a lesser SNR
might be tolerable.
must be selected). This information is commonly stored in the organ and procedure
specific CT protocols and can vary among protocols.
Another innovation in CT that has the potential to significantly reduce doses to
patients is iterative image reconstruction, discussed in Chapter 10. It addresses random noise more effectively than filtered backprojection and can produce images of
equal quality with significantly lower radiation doses.
Two indices of dose to a patient displayed by modern CT scanners are the
CTDIVOL and the DLP, the DLP being the CTDIVOL multiplied by the scanned length
of the patient (see Chapter 11). For scans of a significant length, the CTDIVOL is
approximately the average dose that would be imparted to a cylindrical acrylic
phantom from the scan being given to the patient. Therefore, the CTDIVOL and DLP
provide information about the amount of radiation being given to a specific patient,
but do not reflect the actual doses to the patient.
Modern CT scanners display the CTDIVOL and DLP before and after each scan and
have the ability to send this dose information to a PACS or other digital information
system after a scan, either as a secondary capture image with the study or as a DICOM
Radiation Dose Structured Report (RDSR). If automatic tube current modulation is
used, the CTDIVOL and DLP displayed before a scan may not be identical to those
displayed after the scan. The display of these dose metrics before a scan provides the
technologist the ability to assess the approximate magnitude of the radiation exposure
before performing a scan. The RDSR has a full description of the procedure with radiation dose metrics and includes many details not found in the DICOM metadata associated with the images; however, because the RDSR is a separate DICOM object, reading
876
or storing the information as a RDSR is a separate DICOM transaction. To ensure compatibility in dose reporting, the Integrating the Healthcare Enterprise (IHE) Radiation
Exposure Monitoring integration profile should be specified when purchasing new
imaging systems (IHE integration profiles are discussed in Chapter 5).
CT scanners that comply with the National Electrical Manufacturers Association
(NEMA) XR 25 Computed Tomography Dose Check standard have the capability of
displaying alerts when the CTDIVOL or DLP of a CT study about to be performed is
expected to exceed preset values. A notification value is the dose metric of an individual scan or series of a study that causes a notification warning to be displayed. An
alert value is the cumulative dose metric of the entire study already acquired plus
that for the scan or series about to be performed that causes a warning to be displayed.
By setting notification and alert values, the technologist is warned if the scan he or
she is planning to acquire exceeds the user-configured level of dose. The alert value is
intended to be set to avoid a patient injury and is different from a diagnostic reference
level. The FDA has recommended that the alert value be set at CTDIVOL 5 1 Gy. This
is an unusually high dose for a CT scanabout one-half the level that could result in
skin erythema. If an alert is displayed, the technologist should review the acquisition
parameters for an error. If the alert cannot be resolved, radiologist approval should
be obtained before proceeding, except in a dire emergency. The notification level
could be set to the DRL level. However that would result in numerous notifications,
particularly for obese patients. The American Association of Physicists in Medicine
(AAPM) has recommended that the notification level be set higher than the DRL so that
notifications are issued only for doses that are higher than usual, thereby warning the
technologist that an acquisition parameter may be set incorrectly. The following AAPM
recommended notification values are shown below, (AAPM 2011). This document also
contains recommendations for notification levels for pediatric CT examinations.
Examination
CTDIVOL (mGy)
Adult head
80
Adult torso
50
Brain perfusion
600
Cardiacretrospective gating
150
Cardiacprospective gating
50
Dose Optimization in CT
The key to optimizing dose in CT imaging is to review and optimize the CT protocols.
This is a substantial and complicated task and an overview will be provided herein.
Each protocol should be reviewed from several perspectives. Many scans consist of a
noncontrast scan and one or more scans after the administration of contrast material.
Eliminating one or more phases from a protocol will greatly reduce patient dose.
Thin
slices are usually noisier and in some cases may require a higher mA. One
should consider whether high spatial resolution is diagnostically necessary. One
might overcome the noise of thin slices by calculating thicker slices but at intervals
smaller than the slice thickness (i.e., overlapping). Thin slices exhibit less partial
volume effect and so benefit from higher tissue contrast which could allow dose to
be reduced through lower mAs (ICRP, 2007).
Tissues with inherently high tissue contrast may be acquired at low dose without
loss of diagnostic value. These include lung and bone. A renal stone protocol, for
example, could be acquired at lower dose than a liver protocol.
877
Patients
who are obese generally require greater dose. However, a factor that balances this effect is that adipose tissue better delineates soft tissues, which causes
acceptable contrast to be more easily achieved in obese than in thin patients. A kV
of 140 could be selected for the most obese patients.
Create separate abdominal protocols for obese patients.
Lower kV can increase the visibility of contrast material in angiography or brain
perfusion imaging. The k-absorption edge of iodinewhere its attenuation coefficient reaches a maximumis at about 30 keV. Improved images and lower patient
dose may be achieved with a tube voltage of 100 kV instead of 120 kV in angiography. A tube voltage of 80 kV is commonly used in brain perfusion imaging.
Scan only the z-axis length you need.
When possible, avoid radiation directly to the eyes.
Review the clinical necessity of multiple contrast phases.
Only scan the contrast phases necessary for the diagnostic task. For example, follow-up exams may not require precontrast images.
Vary tube current according to contrast phase. You may be able to use a lower mA
for noncontrast phases.
Compare your protocols with optimized protocols for the same model of CT.
Sample CT protocols are available from the CT manufacturer, AAPM, and other
sources. (See the Image Wisely Web site.)
Look to see that protocols of similar body parts have similar dose indices, taking
into consideration that some indications can tolerate less relative noise.
Use iterative reconstruction, if available.
Use automatic tube current modulation (see below) for procedures for which it will
reduce dose. Most procedures should use it. (Not all procedures, e.g., brain perfusion imaging, benefit from it.)
Ensure body imaging protocols that do not use automatic mA modulation require
technique adjustment for patient body habitus.
Create low-dose protocols (e.g., low-dose chest protocol, low-dose kidney stone
protocol, low-dose AAA protocol) for use when indicated, particularly for follow-up
examinations.
Most automatic mA modulation systems have a maximum allowable mA setting.
Ensure this is set to a reasonable value, beyond which you increase the kV.
Ensure the CTDIVOL for brain perfusion CT does not exceed 0.5 Gy.
Record the CTDIVOL and DLP for a patient of average size in each protocol document. You may need to average the values from several patients.
Compare these dose indices with DRLs.
For cardiac CT protocols, significant reduction in dose can be achieved by the following strategies:
Minimize
the scan range.

heart rate reduction.
Prospective triggering results in lower dose than does retrospective gating.
If one is using retrospective gating, employ ECG-gated mA modulation.
Reduce tube voltage whenever possible to 100 kV or even 80 kV.
Use
Miscellaneous Considerations
Careful identification of the patient and, for female patients of reproductive age,
determination of pregnancy status are necessary before an examination is performed.
For higher-dose studies (e.g., fluoroscopy and multiphase CT examinations) of the
abdomen or pelvis, a pregnancy test within 72 h before the examination should be
performed in addition to the usual screening for pregnancy, unless the possibility of
878
pregnancy is eliminated by certain premenarche in a child, certain postmenopausal

state, or documented hysterectomy or tubal ligation.
A significant reduction in population dose can be achieved by eliminating
screening exams that only rarely detect pathology. Standing-order x-rays, such as
presurgical chest exams for hospital admissions or surgeries not involving the chest,
are inappropriate. Frequent screening examinations are often not indicated without a specific reason or a reasonable amount of time elapsed (e.g., years) from the
previous exam. A good example is the yearly dental x-ray. While yearly exams
are appropriate for some patients, a 2-year interval between x-ray examinations
in noncavity-prone adult patients with a healthy mouth is appropriate. Another
opportunity for significant patient dose reduction in dentistry is the use of highspeed film (e.g., E-speed film) or digital image receptors. Many dental offices still
use slower D-speed films that require approximately twice the exposure (and thus
patient dose). Periodic screening mammography examinations are not appropriate
for women younger than 35 to 40 years old, unless there is a familial history of
breast cancer or other indication. However, as the probability of cancer increases
with age, and the fact that survival is drastically enhanced by early detection makes
yearly screening mammograms sensible for the general population of women 40 to
50 years old.
Each radiology department should have a program to monitor examinations that
must be repeated. The frequency of repeat exams due to faulty technique or improper
position is primarily determined by the skill and diligence of the x-ray technologists.
Training institutions often have higher repeat rates, mainly attributed to lack of experience. The largest numbers of studies that must be repeated are from mobile exams,
in which positioning difficulty causes the anatomy of interest to be improperly represented on the image, and lack of automatic exposure control increases the likelihood
of improper optical densities or dose to the image receptor.
Programmed techniques for examinations are commonly available on radiographic equipment, and can eliminate the guesswork in many radiographic situations (of course, provided that the techniques are properly set up). Technique charts
for various examinations should be posted conspicuously at the control panel to aid
the technologist in the correct selection of radiographic technique. The use of photostimulable phosphor imaging plates or direct digital image receptors (Chapter 7) can
significantly reduce the number of exams that must be repeated because of improper
radiographic technique. The examinations with the highest retake rates are portable
chests; lumbar spines; thoracic spines; kidneys, ureters, and bladders; and abdomens.
Continuous monitoring of retakes and inadequate images and identification of their
causes are important components of a quality assurance program, so that appropriate
action may be taken to improve quality.
Equipment problems, such as improperly loaded cassettes, excessive fog due
to light leaks or poor film storage conditions, processor artifacts caused by dirty
components or contaminated chemicals, uncalibrated x-ray equipment, or improper
imaging techniques can also lead to retakes. Many of these errors are eliminated
by a quality control program that periodically tests the performance of the x-ray
equipment, image receptors, and processing systems.
Reference Levels
A reference level, also known as a diagnostic reference level, is a particular value of a
metric of dose to a patient or a phantom, such as entrance air kerma in radiography
or CTDIVOL in CT, to be used in a quality assurance program to identify possibly
879
excessive doses to patients or protocols or equipment that may be imparting excessive doses. A dose value from an imaging procedure that exceeds a reference level
should be investigated, but would not be presumed as excessive until so determined
by an investigation. Reference levels were mentioned in Chapter 11. In the case
of fluoroscopically guided interventional procedures, comparing dose metrics from
actual clinical procedures to reference levels may help to identify clinicians who are
not making full use of the dose-optimizing features of the imaging devices or who are
less adept at the clinical procedures.
Reference levels may be based upon measurements made with a phantom, such as
CTDI measurements made using the FDA CT body or head phantom, or they may be
based upon dose metrics, such as CTDIVOL and DLP, from clinical imaging of patients.
There are advantages and disadvantages, discussed below, to each approach.
A reference level is typically determined from a study of a dose metric from many
institutions and imaging systems. A reference level is typically set at the seventy-fifth
percentile of the set of data from the study, meaning that three fourths of the measurements in the data set were less than the reference level.
In the United States, the Nationwide Evaluation of X-ray Trends (NEXT) Program, conducted jointly by the US Food and Drug Administration, the Conference
of Radiation Control Program Directors, and most state radiation regulatory agencies,
periodically performs measurements of doses from x-ray imaging devices. NEXT
studies have been of phantom-based dose measurements. The American College of
Radiology (ACR) publishes reference levels determined from dose data from its imaging modality accreditation program. The ACRs reference levels are also based upon
measurements using phantoms. A possible limitation of the ACRs reference levels
is that they are determined only from dose measurements from imaging facilities
seeking accreditation. European governmental agencies are also a source of reference
levels. Reference levels may also be provided by studies reported in the professional
literature (e.g., Hausleiter et al., 2009).
There are shortcomings to the use of reference levels as a quality assurance tool.
Reference levels for dose measurements made using phantoms may not fully test an
imaging device and imaging protocol. For example, the cylindrical FDA CT dose
phantoms have shapes very different from patients and do not assess the effect of
automatic tube current modulation. For another example, a measurement of a fluoroscope with a phantom does not assess the clinicians knowledge in the proper use
of the imaging system or his or her skill in the clinical procedure. On the other hand,
using dose measurements and reference levels determined from actual patient studies involves uncertainty due to factors such as patient body size and, in the case of
fluoroscopically guided interventional procedures, the difficulty of individual patient
procedures. The studies used to determine reference levels may lag current technology and uses; furthermore, because the studies used to determine reference levels are
costly of resources, there are many clinical imaging protocols for which there are no
appropriate reference levels. Lastly, comparisons of dose metrics to reference levels
identify high outliers; a dose metric that is less than a reference level does not imply
that the protocol or procedure is fully optimized.
Nonetheless, reference levels are a useful tool in identifying imaging protocols,
equipment, and practices imparting unnecessarily high doses to patients. As mentioned in Chapter 11, the automated collection of dose information, using standard
DICOM and IHE protocols, from clinical procedures and its collection by dose registries may facilitate the timely creation and updating of reference levels and the
comparison of dose metrics from clinical procedures with reference levels.
880
21.7 Radiation Protection in Nuclear Medicine
Minimizing Time of Exposure to Radiation Sources

Nuclear medicine procedures typically produce low exposure rates for extended
periods of time. For example, the typical exposure rate at 1 m from a patient after the
injection of approximately 740 MBq (20 mCi) of technetium 99m methylenediphosphonate (Tc-99m MDP) for a bone scan is approximately 1 mR/h, which, through
radioactive decay and urinary excretion, will be reduced to approximately 0.5 mR/h
at 2 h after injection, when imaging typically begins. Therefore, knowledge of both
the exposure rate and how it changes with time are important in minimizing personnel exposure. Table 21-5 shows the typical exposure rates at 1 m from adult nuclear
medicine patients after the administration of commonly used radiopharmaceuticals.
The time spent near a radiation source can be minimized by having a thorough
understanding of the tasks to be performed and the appropriate equipment to complete them in a safe and timely manner. For example, elution of a Mo-99/Tc-99m
generator and subsequent radiopharmaceutical preparation requires several steps
with a high-activity source. It is important to have practiced these steps with a nonradioactive source to learn how to manipulate the source proficiently and use the
dose measurement and preparation apparatus.
Minimizing Exposure with Distance from Radiation Sources

Other than low-activity sealed sources (e.g., a 370-kBq [10 Ci] Co 57 marker),
unshielded radiation sources should never be manipulated by hand. The use of
tongs or other handling devices to increase the distance between the source and
the hand substantially reduces the exposure rate. Table 21-6 lists the exposure rates
from radionuclides commonly used in nuclear medicine and the 100-fold exposure
rate reduction achieved simply by increasing the distance from the source from 1 to
10cm.
Radiation exposure rates in the nuclear medicine laboratory can range from over
100 R/h (e.g., contact exposure from an unshielded vial containing approximately
37 GBq [1 Ci] of Tc-99m generator eluate or a therapeutic dosage of approximately
TABLE 21-5TYPICAL EXPOSURE RATE AT 1 meter FROM an ADULT
NUCLEAR MEDICINE PATIENT AFTER RADIOPHARMACEUTICAL

ADMINISTRATION
RADIOPHARMACEUTICAL
ACTIVITY
MBq (mCi)
EXPOSURE RATE AT 1 m
mGy/h (mR/h)
Thyroid cancer therapy
I-131 (NaI)
7,400 (200)
~306 (35)
Tumor imaging
F-18 (FDG)
370 (10)
~43.7 (5.0)
Cardiac-gated imaging
Tc-99m RBC
740 (20)
~7.9 (0.9)
Bone scan
Tc-99m MDP
925 (25)
~10.5 (1.2)
Tumor imaging
Go-67 citrate
111 (3)
~3.5 (0.4)
Liver-spleen scan
Tc-99m sulfur colloid
148 (4)
~1.8 (0.2)
Myocardial perfusion
imaging
Tl-201 chloride
111 (3)
~0.87 (0.1)
STUDY
Exposure rate measurements courtesy of M. Hartman, University of California, Davis.
881
TABLE 21-6EFFECT OF DISTANCE ON EXPOSURE WITH COMMON
RADIONUCLIDES USED IN NUCLEAR MEDICINE

RADIONUCLIDE 370 MBq (10 mCi)
EXPOSURE RATEa mGy/hb

(R/h) AT 1 cm
EXPOSURE RATEa mGy/hb

(R/h) AT 10 cm (~4 inches)
Ga-67
65.5 (7.5)
0.65 (0.075)
Tc-99m
54.1 (6.2)
0.54 (0.062)
I-123
143 (16.3)
1.43 (0.163)
I-131
190 (21.8)
1.90 (0.218)
Xe-133
46.3 (5.3)
0.46 (0.053)
Tl-201
39.3 (4.5)
0.39 (0.045)
F-18
494 (56.6)
4.9 (0.57)
Calculated from the 20.

Air kerma rate.
11 GBq [300 mCi] of I-131) to natural background. The air kerma rate at any distance from a small container of a particular radionuclide can be calculated using the
air-kerma-rate constant (Gd), of the radionuclide. The air-kerma-rate constant for a
photon-emitting radionuclide is defined by the ICRU as the quotient of l2 K& d by A,
where K& d is the air kerma rate due to photons of energy greater than , at a distance
l from a point source of the nuclide having an activity A,

d =
l2 K& d
A
[21-4]
The air-kerma-rate constant has units of m2 Gy Bq1 s1. For convenience, the airkerma-rate constant is sometimes expressed in m2 Gy GBq1 h1.
Because very low energy photons are significantly attenuated in air and other intervening materials, air kerma rate constants usually ignore photons below a particular
energy. For example, G20 represents the air-kerma-rate constant for photons 20 keV.
Example: An unshielded vial containing 3.7 GBq of Tc-99m pertechnetate is left on a lab
bench. For Tc-99m, G20 = 14.6 m2GyGBq21h2l What would be the air kerma (K air ) if a technologist were standing 0.5 m away for 30 min?
Solution:
2
K& air = (14.6 m2 Gy GBq 1 h l ) (3.7 GBq ) (0.5 h ) / (0.5 m ) = 108 Gy
Multiplying the air kerma in Gy by 0.1145 mR/Gy will convert it to exposure

in units of mR; thus 108 Gy is equivalent to (108 Gy 3 0.1145 mR/Gy) =
12.4 mR.
The air-kerma-rate constant has largely replaced the older exposure rate
constant (expressed in units of Rcm2mCi1h1) where the exposure rate in R/h at
1 cm from 1 mCi of the specified radionuclide was expressed as

Exposure rate (R/h) = GxA/d2
[21-5]
where Gx is the exposure rate constant Rcm2mCi1h1 for photons of energy greater
than x, A is the activity in mCi, and d is the distance in centimeters from a point
source of radioactivity.
882
One can multiply the exposure rate constant in Rcm2mCi1h1 by 21.69 to convert to air-kerma-rate constant in GyGBq1h1. This conversion assumes that the
photon energy cutoff (i.e., and x) is the same.
Shielding in Nuclear Medicine

Tungsten, lead, leaded glass, and leaded acrylic shields are used in nuclear medicine to reduce the radiation exposure from vials and syringes containing radioactive
material. Table 21-7 shows exposure rate constants and lead HVLs for radionuclides
commonly used in nuclear medicine.
Syringe shields (Fig. 21-18) are used to reduce personnel exposure from syringes
containing radioactivity during dose preparation and administration to patients.
Syringe shields can reduce hand exposure from Tc-99m by as much as 100-fold.
Leaded glass or acrylic shields are used in conjunction with solid lead shields in
radiopharmaceutical preparation areas. Radiopharmaceuticals are withdrawn from
vials surrounded by thick lead containers (called lead pigs) into shielded syringes
behind the leaded glass shield in the dose preparation area (Fig. 21-19).
Persons handling radionuclides should wear laboratory coats, disposable gloves,
finger ring TLD dosimeters, and body dosimeters. The lead aprons utilized in
diagnostic radiology are of limited value in nuclear medicine because, in contrast to
their effectiveness in reducing exposure from low-energy scattered x-rays, they do
not attenuate enough of the medium-energy photons emitted by Tc-99m (140 keV)
to be practical (Table 21-8).
Radioactive material storage areas are shielded to minimize exposure rates. Mirrors mounted on the back walls of high-level radioactive material storage areas are
often used to allow retrieval and manipulation of sources without direct exposure to
the head and neck. Beta radiation is best shielded by low atomic number (Z) material (e.g., plastic or glass), which provides significant attenuation while minimizing bremsstrahlung x-ray production. For high-energy beta emitters (e.g., P-32), the
low Z shield can be further shielded by lead to attenuate bremsstrahlung. With the
exception of PET facilities, which often shield surrounding areas from the 511-keV
annihilation radiation, most nuclear medicine laboratories do not find it necessary to
shield the walls within or surrounding the department.
FIGURE 21-18 A butterfly needle

and three-way stopcock connected
to a syringe with normal saline
and a syringe shield containing the
syringe with the radiopharmaceutical for injection. The barrel of the
syringe shield is made of a high-Z
material (e.g., lead or tungsten)
in which a leaded glass window is
inserted so that the syringe scale
and the contents of the syringe can
be seen.
883
TABLE 21-7EXPOSURE RATE CONSTANTS (G20 AND G30)a AND HALF VALUE
LAYERS (HVL) OF LEAD FOR RADIONUCLIDES OF INTEREST TO

NUCLEAR MEDICINE
RADIONUCLIDE
G20b,c
C-11, N-13, O-15
139 (5.85)
Co-57
13.2 (0.56)
Co-60
303 (12.87)
D20d
G30c
HVL IN Pb (cm)e,f
153.10
5.85
0.39
14.51
0.56
0.02
333.41
12.87
1.2
Cr-51
4.24 (0.18)
4.66
0.18
0.17
Cs-137/Ba-137m
76.5 (3.25)
84.19
3.25
0.55
F-18
135 (5.66)
148.44
5.66
0.39
Ga-67
17.7 (0.75)
19.43
0.75
0.1
I-123
38.4 (1.63)
42.23
0.86
0.04
1-125
34.6 (1.47)
38.08
0.26
0.002
I-131
51.3 (2.18)
56.47
2.15
0.3
In-111
75.4 (3.20)
82.90
2.0
0.1
Ir-192
108.6 (4.61)
119.43
4.61
0.60
38.08
1.43
0.7
5.65
0.7
Mo-99/Tc-99m
34.6 (1.47)
Rb-82
134 (5.65)
Tc-99m
14.6 (0.62)
16.06
0.60
0.03
Tl-201
10.6 (0.45)
11.66
0.45
0.02
Xe-133
12.5 (0.53)
13.73
0.53
0.02
159
a
(G20 and G30) calculated from the absorption coefficients of Hubbell and Seltzer (1995) and the decay data
table of Kocher (1981): Hubbell JH, Seltzer SM. Tables of x-ray mass attenuation coefficients and mass energyabsorption coefficients 1 keV to 20 MeV for elements Z = 1 to 92 and 48 additional substances of dosimetric
interest. NISTIR 5632, National Institute of Standards and Technology, May 1995 and Kocher DC. Radioactive
decay data tables. DOE/TIC-11026, Technical Information Center, U.S. Department of Energy, 1981.
b
Air kerma rate constant, 20 keV photon energy cutoff, in mGy m2 GBq1 h1.
c
(Rcm2/mCih): Multiply by 23.69 to obtain air kerma rate constant in mGym2/GBqh.
d
Equivalent dose rate mSv m2 GBq1 h1.
e
The first HVL will be significantly smaller than subsequent HVLs for those radionuclides with multiple photon
emissions at significantly different energies (e.g., Ga-67) because the lower-energy photons will be preferentially attenuated in the first HVL.
f
Some values were adapted from Goodwin PN. Radiation safety for patients and personnel. In: Freeman and
Johnsons clinical radionuclide imaging. 3rd ed. Philadelphia, PA: WB Saunders, 1984:370. Other values were
calculated by the authors.
g
In equilibrium with Tc-99m.
Contamination Control and Surveys

Contamination is simply uncontained radioactive material located where it is not
wanted. Contamination control methods are designed to prevent radioactive material from coming into contact with people and to prevent its spread to other work
surfaces. Protective clothing and handling precautions to control contamination are
similar to the universal precautions that are taken to protect hospital personnel from
pathogens. In most cases, disposable plastic gloves, laboratory coats, and closed-toe
shoes offer adequate protection. One advantage of working with radioactive material,
in comparison to other hazardous substances, is that small quantities can be easily
detected. Personnel and work surfaces must be routinely surveyed for contamination.
Nonradioactive work areas near radioactive material use areas (e.g., reading rooms
and patient exam areas) should be posted as clean areas where radioactive materials
884
FIGURE 21-19 A. Dose preparation workstation. The technologist

is placing a syringe containing a
radiopharmaceutical into a syringe
shield. Protection from radiation
exposure is afforded by working
behind the lead L-shield with a
1.8-mm lead equivalent glass window that reduces the radiation
exposure from Tc-99m by approximately 100. The technologist is
wearing a lab coat and disposable
gloves to protect his skin and clothing from contamination. A film
badge is worn on the lab coat to
record body exposure and a TLD
finger ring dosimeter is worn inside
the glove with the TLD chip facing
the source to record the extremity exposure. B. A much thicker
syringe shield is required for PET
radiopharmaceuticals to attenuate the 511 keV annihilation photons (left) compared to the typical
syringe shields used for shielding
the 140 keV photons from Tc-99m
(right).
TABLE 21-8EXPOSURE REDUCTION ACHIEVED BY A 0.5-mm LEAD
EQUIVALENT APRON FOR VARIOUS RADIATION SOURCES

SOURCE
EXPOSURE REDUCTION NO. OF APRONS TO REDUCE

ENERGY (keV) WITH 1 APRON
EXPOSURE BY 90% (I.E., 1 TVL) WEIGHT (lb)
Scattered x-rays 1030
.90%
,1
,15
Tc-99m
140
,70%
,2
,30
Cs-137
662
,6%
,36
,540
885
are prohibited. Work surfaces where unsealed radioactive materials are used should
be covered by plastic-backed absorbent paper that is changed when contaminated or
worn. Volatile radionuclides (e.g., I-131 and Xe-133 gas) should be stored in a fume
hood with 100% exhaust to the exterior of the building to prevent airborne contamination and subsequent inhalation. Personnel should discard gloves into designated
radioactive waste receptacles after working with radioactive material and monitor their
hands, shoes, and clothing for contamination periodically. All personnel should wash
their hands after handling radioactive materials and especially before eating or drinking to minimize the potential for internal contamination. If the skin becomes contaminated, the best method of decontamination is washing with soap and warm water. The
skin should not be decontaminated too aggressively to avoid creating abrasions that
can enhance internal absorption. External contamination is rarely a serious health hazard; however, internal contamination can lead to significant radiation exposures. Good
contamination control techniques help prevent inadvertent internal contamination.
The effectiveness of contamination control is monitored by periodic GM meter
surveys (after handling unsealed radioactive material, whenever contamination is
suspected, and at the end of each workday) and wipe tests (typically weekly) of radionuclide use areas. Wipe tests are performed by wiping surfaces using small pieces
of filter paper or cotton-tipped swabs to check for removable contamination at various locations throughout the nuclear medicine laboratory. These wipe test samples
(called swipes) are counted in a NaI (Tl) gamma well counter. Areas that are demonstrated to be in excess of twice background radiation levels are typically considered to be contaminated. The contaminated areas are immediately decontaminated
followed by additional wipe tests to confirm the effectiveness of decontamination
efforts. GM meter surveys are also performed throughout the department at the end
of the day to detect areas of contamination. In addition, exposure rate measurements
(with a portable ion chamber survey meter) are made near areas that could produce
high exposure rates (e.g., radioactive waste and material storage areas).
Laboratory Safety Practices

The following are other laboratory safety practices that minimize the potential for
personnel or facility contamination. Examples of these radiation safety practices are
shown in Figure 21-20.
Label
each container of radioactive material with a Caution Radioactive Material label and the radionuclide, measurement or calibration date, activity, and
chemical form.
Wear laboratory coats or other protective clothing at all times in areas where
radioactive materials are used.
Wear disposable gloves at all times while handling radioactive material.
Either after each procedure or before leaving the area, monitor hands and clothing for
contamination in a low-background area using an appropriate survey instrument.
Use syringe shields for reconstitution of radiopharmaceutical kits and administration of radiopharmaceuticals to patients, except when their use is contraindicated
(e.g., recessed veins, infants). In these and other exceptional cases, use other protective methods, such as remote delivery of the dose (e.g., use a butterfly needle).
Wear personnel monitoring devices, if required, at all times while in areas where radioactive materials are used or stored. When not being worn to monitor occupational
exposures, personnel monitoring devices shall be stored in the work place in a
designated low-background area.
Wear extremity dosimeters, if required, when handling radioactive material.
886
FIGURE 21-20 Routine radiation safety

practices in nuclear medicine. A. Disposal of
radioactive waste into a shielded container
and proper posting of caution radioactive
material signs. B. Unit dosages of radiopharmaceuticals come from a commercial
radiopharmacy in lead shielded tubes that
are cushioned for transport. C. Package survey with ion chamber survey meter to verify
the shielding of the radioactive material is
intact. D. Package labels state the radionuclides, the activity, and the maximum exposure rates expected at one meter from the
package surface.
887
Dispose
of radioactive waste only in designated, labeled, and properly shielded

receptacles.
Shield sources of radioactive material to reduce potential exposure to people.
Perform package receipt survey to assure the source shielding has not been damaged in transport and that the package is not leaking or contaminated.
Do not eat, store food, drink, smoke, or apply cosmetics in any area where radioactive material is stored or used.
Perform Xe-133 ventilation studies in a room with negative pressure with respect to
the hallway. (Negative pressure will occur if the room air exhaust rate substantially
exceeds the supply rate. The difference is made up by air flowing from the hallway
into the room, preventing the escape of the xenon gas.) Xe-133 exhaled by patients
undergoing ventilation studies must be exhausted into a Xe trap. Xe traps have large
shielded activated charcoal cartridges, which adsorb xenon on the large surface
areas presented by the charcoal, thus slowing its migration through the cartridge
and allowing significant decay to occur before it is exhausted to the environment.
Report serious spills or accidents to the radiation safety officer (RSO) or health
physics staff.
Radioactive Material Spills

Urgent first aid takes priority over decontamination after an accident, and personnel
decontamination takes priority over facility decontamination efforts. Individuals in
the immediate area should be alerted to a spill and remain until they can be monitored with a GM survey instrument to ensure that they have not been contaminated.
Radioactive material spills should be contained with absorbent material and the area
isolated and posted with warning signs indicating the presence of radioactive contamination. Disposable shoe covers and plastic gloves should be donned before beginning decontamination. Decontamination should be performed from the perimeter of
the spill toward the center to limit the spread of contamination. Decontamination is
usually accomplished by simply absorbing the spill and cleaning the affected areas
with detergent and water. A meter survey and wipe test should be used to verify successful decontamination. Personnel participating in decontamination should remove
their protective clothing and be surveyed with a GM survey instrument to assure that
they are not contaminated. If the spill involves volatile radionuclides or if other conditions exist that suggest the potential for internal contamination, bioassays should
be performed by the radiation safety staff. Common bioassays for internal contamination include external counting with a NaI (Tl) detector (e.g., thyroid bioassay for
radioiodine) and measurement of radioactivity in urine (e.g., tritium bioassay).
Protection of the Patient in Nuclear Medicine

Sometimes nuclear medicine patients are administered the wrong radiopharmaceutical or the wrong activity. Depending upon the activity, dose, and other specifics,
these accidents may meet the regulatory criteria for what are referred to as
medical events (see Chapter 16) which have prescriptive reporting and quality
improvement requirements. The following event has occurred at several institutions: a patient referred for a whole-body bone scan (typically 740 MBq [20 mCi]
of Tc-99m MDP) has been mistakenly administered up to 370 MBq (10 mCi) of
I-131 sodium iodide for a whole-body thyroid cancer survey [thyroidal dose for
370 MBq and 20% uptake is ~100 Gy (10,000 rad)]. The cause has often been a
verbal miscommunication. In another case, a mother who was nursing an infant was
administered a therapeutic dose of I-131 sodium iodide without being instructed to
888
discontinue breast-feeding, resulting in an estimated dose of 300 Gy (30,000 rad) to

the infants thyroid. The following precautions, most of which are mandated by the
U.S. Nuclear Regulatory Commission (NRC), are intended to reduce the frequency
of such incidents.
Each syringe or vial that contains a radiopharmaceutical must be conspicuously
labeled with the name of the radiopharmaceutical or its abbreviation. Each syringe
or vial shield must also be conspicuously labeled, unless the label on the syringe or
vial is visible. It is prudent to also label each syringe or vial containing a dosage of a
radiopharmaceutical with the patients name. The patients identity must be verified,
whenever possible by two means (e.g., by having the patient recite his or her name
and birth date). The possibility that a female adolescent or a woman is pregnant or
nursing an infant by breast must be ascertained before administering the radiopharmaceutical. In the case of activities of I-131 sodium iodide exceeding 30 Ci and
therapeutic radiopharmaceuticals, pregnancy should be ruled out by a pregnancy
test within 72 h before the procedure, certain premenarche in a child, certain postmenopausal state, or documented hysterectomy or tubal ligation.
Before the administration of activities exceeding 1.1 MBq (30 Ci) of I-131 in
the form of sodium iodide and any radionuclide therapy, a written directive must
be prepared by the nuclear medicine physician identifying the patient, the radionuclide, the radiopharmaceutical, the activity to be administered, and the route of
administration. The written directive must be consulted at the time of administration and the patients identity should be verified by two methods. Although not
required by the NRC, similar precautions should be taken when reinjecting autologous radiolabeled blood products to prevent transfusion reactions and the transmission of pathogens.
In some cases, women who are nursing infants by breast at the time of the examination may need to be counseled to discontinue breast-feeding until the radioactivity in the
breast milk has been reduced to a safe level. Table 21-9 contains recommendations for
cessation of breast-feeding after the administration of radiopharmaceuticals to mothers.
Radiation doses from the activity ingested by a nursing infant have been estimated
for the most common radiopharmaceuticals used in diagnostic nuclear medicine
(Stabin and Breitz, 2000). In many cases, no interruption in breast-feeding is needed
to maintain a radiation dose to the infant well below 1 mSv. Only brief interruption
(hours to days) of breast-feeding was advised for other Tc-99m radiopharmaceuticals:
macroaggregated albumin, pertechnetate, red blood cells, white blood cells, as well
as I-123 metaiodobenzylguanidine, and Tl-201. Complete cessation is suggested for
Ga-67 citrate and I-131, however prior recommendation for cessation following I-123
NaI administration were based on a 2.5% contamination with I-125, which is no longer applicable (Siegel, 2002).
The NRC requires written instructions be given to nursing mothers if the total
effective dose equivalent (TEDE) to a nursing infant could exceed 1 mSv (Code of
Federal Regulations Title 10 Part 35.75). The instructions include guidance on the
interruption or discontinuation of breast-feeding, and information on the potential
consequences, if any, of failure to follow the guidance.
Radionuclide Therapy
Treatment of thyroid cancer and hyperthyroidism with I-131 sodium iodide is a
proven and widely utilized form of radionuclide therapy. NaI-131 is manufactured in
liquid and capsule form. I-131 in capsules is incorporated into a waxy or crystalline
matrix or bound to a gelatin substance, all of which reduce the volatility of the
I-131. In recent years, advances in monoclonal antibody and chelation chemistry
889
TABLE 21-9RECOMMENDED DURATION OF INTERRUPTION OF
BREAST-FEEDING FOLLOWING RADIOPHARMACEUTICAL

ADMINISTRATION TO A PATIENT WHO IS NURSING AN INFANT
OR CHILD
RADIOPHARMACEUTICAL
ADMINISTERED
ACTIVITY MBq (mCi)
DURATION OF INTERRUPTION
OF BREAST-FEEDINGa
Tc-99m DTPA, MDP, PYP, RBC, or GH
740 (20)
None
Tc-99m sestamibi or tetrofosmin
1,110 (30)
None
Tc-99m disofenin
300 (8)
None
Tc-99m MAG3
370 (10)
None
Tc-99m sulfur colloid
444 (12)
None
I-123OIH
74 (2)
None
Tc-99m sodium pertechnetate
185 (5)
4h
Tc-99m MAA
148 (4)
12 h
F-18 FDGb
740 (20)
12 hb
I-123 NaI
14.8 (0.4)
24 h
I-123 MIBG
370 (10)
48 h
Tc-99m WBC
185 (5)
48 h
In-111 WBC
18.5 (0.5)
1 wk
Tl-201 chloride
111 (3)
96 h
Ga-67 citrate
185 (5)
Discontinuec
I-131 NaI
1 (0.027)
Discontinuec
None means that interruption of breast-feeding need not be recommended, given criterion of a limit of
1 mSv effective dose to infant and these administered activities. However, even this low dose may be reduced
by a 12 to 24-h interruption of breast-feeding.
a
Adapted from NUREG 1556, Vol. 9, Rev. 2, Program-Specific Guidance About Medical Use Licenses,
Appendix U, U.S. Nuclear Regulatory Commission, January 2008. (See Romney BM, Nickloff EL, Esser PD, et al.
Radionuclide administration to nursing mothers: mathematically derived guidelines. Radiology 1986;160:
549554; for derivation of milk concentration values for radiopharmaceuticals.)
b
Minimal F-18 FDG in breast milk (J Nucl Med 2001;42(8):12381242). Waiting 6 half-lives (12 h) lowers the
exposure to the infant from the mother.
c
Discontinuance is based not only on the long duration recommended for cessation of breast-feeding but also
on the high dose the breasts themselves would receive during the radiopharmaceutical breast transit.
Stabin MG, Breitz HB. Breast milk excretion of radiopharmaceuticals: mechanisms, findings, and radiation
dosimetry. J Nucl Med 2000;41(5):863873.
DTPA, diethylenetriaminepentaaceticacid; MAA, macroaggregated albumin; disofenin is an iminodiacetic
acid derivative; FDG, fluorodeoxyglucose; MIBI, methoxyisobutylisonitrile; MDP, methylene diphosphonate;
PYP, pyrophosphate; RBC, red blood cells; WBC, white blood cells; OIH, orthoiodohippurate; MIBG,
metaiodobenzylguanidine; GH, glucoheptonate; MAG3, mercaptoacetyltriglycine.
t echnology have produced a variety of radioimmunotherapeutic agents, many labeled

with I-131. I-131 decays with an 8-day half-life and emits high-energy beta particles
and gamma rays. These decay properties, together with the facts that I-131 can be
released as a gas under certain conditions, can be absorbed through the skin, and
concentrates and is retained in the thyroid, necessitate a number of radiation protection precautions with the administration of therapeutic quantities of I-131.
Following administration, I-131 is secreted or excreted in all body fluids including
urine, saliva, and perspiration. In some cases the patient may need to be hospitalized
as a radiation safety precaution or because of comorbidities (criteria for hospitalization
are discussed below). If a patient is to be hospitalized, before I-131 is administered,
surfaces of the patients room likely to become contaminated, such as the floor, bed
890
controls, mattress, light switches, toilet, and telephone, are covered with plastic or
absorbent plastic-backed paper to prevent contamination. In addition, the patients
meals are served on disposable trays. Containers are placed in the patients room to
dispose of used meal trays and to hold contaminated linens for decay. Radiation safety
staff will measure exposure rates at the bedside, at the doorway, and in neighboring
rooms. In some cases, immediately adjacent rooms must be posted off limits to control radiation exposure to other patients and nursing staff. These measurements are
posted, together with instructions to the nurses and visitors including maximal permissible visiting times. Nursing staff are required to wear dosimeters and are trained in
the radiation safety precautions necessary to care for these patients safely. Visitors are
required to wear disposable shoe covers, and staff members wear both shoe covers and
disposable gloves to prevent contamination. Visiting times are limited and patients are
instructed to stay in their beds during visits and avoid direct physical contact with
visitors to keep radiation exposure and contamination to a minimum. Visitors and staff
are usually restricted to nonpregnant adults. After the patient is discharged, the health
physics or nuclear medicine staff decontaminates the room and verifies through wipe
tests and GM surveys that the room is sufficiently decontaminated. Federal or state
regulations may require thyroid bioassays of staff technologists and physicians directly
involved with dose preparation or administration of large activities of I-131.
The NRC regulations (Title 10 Part 35.75) require that patients receiving therapeutic radionuclides be hospitalized until or unless it can be demonstrated that the TEDE to
any other individual from exposure to the released patient is not likely to exceed 5 mSv.
Guidance on making this determination can be found in two documents from the NRC:
NUREG-1556, Vol. 9, entitled A Consolidated Guidance about Materials Licenses: Program-Specific Guidance about Medical Licenses and Regulatory Guide 8.39, entitled
Release of Patients Administered Radioactive Materials. These documents describe
methods for calculating doses to other individuals and contain tables of activities not
likely to cause doses exceeding 5 mSv. For example, patients may be released from the
hospital following I-131 therapy when the activity in the patient is at or below 1.2 GBq
(33 mCi) or when the dose rate at 1 m from the patient is at or below 70 Sv/h.
To monitor the activity of I-131 in a patient, an initial exposure measurement is
obtained at 1 m from the patient soon after the administration of the radiopharmaceutical. This exposure rate is proportional to the administered activity. Exposure rate
measurements are repeated daily until the exposure rate associated with 1.2 GBq (33
mCi) is obtained. The exposure rate from 33 mCi in a patient will vary with the mass
of the patient. The activity remaining in the patient can be estimated by comparing initial and subsequent exposure rate measurements made at a fixed, reproducible geometry. The exposure rate equivalent to 1.2 GBq (33 mCi) remaining the patient (X1.2) is
X1.2 =
(1.2 GBq)X 0
A0
where X0 is the initial exposure rate (~15 min after radiopharmaceutical administration) and A0 is the administered activity (GBq).
For example, a thin patient is administered 5.55 GBq (150 mCi) I-131, after
which an exposure rate measurement at 1 m reads 37 mR/h. The patient can be discharged when the exposure rate at 1 m falls below:
X1.2 =
(1.2 GBq) (37 mR/h)

= 8.0 mR/h
5.5 GBq
891
Once below 1.2 GBq (33 mCi), assuming a stable medical condition, the patient may
be released. If the TEDE to any other individual is likely to exceed 1 mSv (0.1 rem),
the NRC requires the licensee to provide the released individual, or the individuals
parent or guardian, with radiation safety instructions. These instructions must be
provided in writing and include recommended actions that would minimize radiation exposure to, and contamination of, other individuals. Typical radiation safety
precautions for home care following radioiodine therapy are shown in Appendix H.
These precautions also apply to patient therapy with less than 1.2 GBq (33 mCi) of
I-131 for hyperthyroidism. On the rare occasions when these patients are hospitalized
for medical or radiation safety reasons (e.g., small children at home), contamination
control procedures similar to thyroid cancer therapy with I-131 should be observed.
Phosphorus-32 as sodium phosphate is used for the treatment of polycythemia vera
and some leukemias, and palliation of pain from cancerous metastases in bone; P-32
in colloidal form is used to treat malignant effusions and other diseases. Strontium-89
chloride and samarium-153 lexidronam are used to treat intractable pain from metastatic
bone disease. The primary radiation safety precaution for these radionuclide therapies
is contamination control. For example, the wound and bandages at the site of an intraabdominal instillation of P-32 should be checked regularly for leakage. The blood and
urine will be contaminated and universal precautions should be observed. Exposure rates
from patients treated with pure beta emitters like P-32 are not significant. Likewise the
exposure rates from patients treated with Sr-89 and Y-90, both of which emit gamma rays
with very low abundances, are insignificant compared to the risk of contamination.
Iodine-131 and yttrium-90, bound to monoclonal antibodies, are used to treat
non-Hodgkins lymphomas that exhibit a particular antigen. The precautions for
therapy with I-131 antibodies are similar to those for therapies using I-131 sodium
iodide. The precautions for therapies with Y-90 antibodies are similar to those in the
previous paragraph.
Radioactive Waste Disposal

Minimizing radioactive waste is an important element of radioactive material use.
Most radionuclides used in nuclear medicine have short half-lives that allow them
to be held until they have decayed. As a rule, radioactive material is held for at least
10 half-lives and then surveyed in an area of low radiation background with an
appropriate radiation detector (typically a GM survey instrument with a pancake
probe) to confirm the absence of any detectable radioactivity before being discarded
as nonradioactive waste. Disposal of decayed radioactive waste in the nuclear medicine department is made easier by segregating the waste into short half-life (e.g., F-18
and Tc-99m), intermediate half-life (e.g., Ga-67, In-111, I-123, Xe-133, Tl-201), and
long half-life (e.g., P-32, Cr-51, Sr-89, I-131) radionuclides.
Small amounts of radioactive material may be disposed of into the sanitary sewer
system if the material is water soluble and the total amount does not exceed regulatory limits. Records of all radioactive material disposals must be maintained for
inspection by regulatory agencies. Radioactive excreta from patients receiving diagnostic or therapeutic radiopharmaceuticals are exempt from these disposal regulations and thus may be disposed into the sanitary sewer. The short half-lives and
large dilution in the sewer system provide a large margin of safety with regard to
environmental contamination and public exposure.
Many solid waste and medical waste disposal facilities have radiation detection
systems to monitor incoming waste shipments. It is prudent for hospitals to take measures to ensure that radioactive patient waste such as diapers are not inadvertently
892
placed into non-radioactive waste containers. Many hospitals have radiation waste
monitoring systems and housekeepers take solid waste containers past them before
discarding the waste into dumpsters.
21.8 Regulatory Agencies and Radiation Exposure Limits
Regulatory Agencies
A number of regulatory agencies have jurisdiction over various aspects of the use of
radiation in medicine. The regulations promulgated under their authority carry the
force of law. These agencies can inspect facilities and records, levy fines, suspend
activities, and issue and revoke radiation use authorizations.
The U.S. NRC regulates special nuclear material (plutonium and uranium
enriched in the isotopes U-233 and U-235), source material (thorium, uranium, and
their ores), and by-product material used in the commercial nuclear power industry,
research, medicine, and a variety of other commercial activities. Under the terms
of the Atomic Energy Act of 1954, which established the Atomic Energy Commission whose regulatory arm now exists as the NRC, the agency was given regulatory
authority only for special nuclear, source, and by-product material. The term byproduct material originally referred primarily to radioactive by-products of nuclear
fission in nuclear reactors, that is, fission products and radioactive materials produced by neutron activation. However, the federal Energy Policy Act of 2005 modified the definition of by-product material to include material made radioactive by
use of a particle accelerator. Thus, accelerator-produced radioactive materials (e.g.,
cyclotron-produced radionuclides such as F-18, Tl-201, I-123, and In-111) are now
subject to NRC regulation.
Most states administer their own radiation control programs for radioactive materials and other radiation sources. These states have entered into agreements with the
NRC, under the Atomic Energy Act of 1954, to promulgate and enforce regulations
similar to those of the NRC and are known as agreement states. These agreement
states conduct the same regulatory oversight, inspection, and enforcement actions
that would otherwise be performed by the NRC. In addition, the agencies of the agreement states typically regulate all sources of ionizing radiation, including diagnostic
and interventional x-ray machines and linear accelerators used in radiation oncology.
While NRC and agreement state regulations are not identical, there are many
essential aspects that are common to all of these regulatory programs. Workers must
be informed of their rights and responsibilities, including the risks inherent in utilizing radiation sources and their responsibility to follow established safety procedures.
The NRCs regulations are contained in Title 10 (Energy) of the Code of Federal
Regulations (CFR). The most important sections for medical use of radionuclides
are the Standards for Protection against Radiation (Part 20) and Medical Use
of By-Product Material (Part 35). Part 20 contains the definitions utilized in
the radiation control regulations and requirements for radiation surveys; personnel monitoring (dosimetry and bioassay); radiation warning signs and symbols; and
shipment, receipt, control, storage, and disposal of radioactive material. Part 20 also
specifies the maximal permissible doses to radiation workers and the public; environmental release limits; and documentation and notification requirements after a
significant radiation accident or event, such as the loss of a brachytherapy source or
the release of a large quantity of radioactive material to the environment.
Part 35 lists the requirements for the medical use of by-product material. Some
of the issues covered in this section include the medical use categories, training
893
r equirements, precautions to be followed in the medical use of radiopharmaceuticals,

testing and use of dose calibrators, and requirements for the reporting of medical
events involving specified errors in the administrations of radiopharmaceuticals to
patients. The NRC also issues regulatory guidance documents, which provide the
licensee with methods acceptable to NRC for satisfying the regulations. The procedures listed in these documents may be adopted completely or in part with the
licensees own procedures, which will be subject to NRC review and approval.
The FDA regulates the development and manufacturing of radiopharmaceuticals
as well as the manufacturing of medical x-ray equipment. Its regulations regarding
the manufacture of medical x-ray imaging equipment contain extensive design and
performance requirements for the equipment. Although this agency does not directly
regulate the end user (except for mammography), it does maintain a strong involvement in both the technical and regulatory aspects of human research with radioactive
materials and other radiation sources and publishes guidance documents in areas of
interest including radiologic health, design and use of x-ray machines, and radiopharmaceutical development. FDA regulations, specifically 21 CFR 803, require the
reporting of serious injuries and deaths that a medical device has or may have caused
or contributed to.
The U.S. Department of Transportation (DOT) regulates the transportation of
radioactive materials. Other regulations and recommendations related to medical
radiation use programs are promulgated by other federal, state, and local agencies.
Advisory Bodies
Several advisory organizations exist that periodically review the scientific literature
and issue recommendations regarding various aspects of radiation protection. While
their recommendations do not constitute regulations and thus do not carry the
force of law, they are usually the origin of most of regulations adopted by regulatory
agencies and are widely recognized as standards of good practice. Many of these
recommendations are voluntarily adopted by the medical community even in the
absence of a specific legal requirement. The two most widely recognized advisory
bodies are the National Council on Radiation Protection and Measurements (NCRP)
and the International Commission on Radiological Protection (ICRP). The NCRP
is a nonprofit corporation chartered by Congress to collect, analyze, develop, and
disseminate, in the public interest, information and recommendations about radiation protection, radiation measurements, quantities, and units. In addition, it is
charged with working to stimulate cooperation and effective utilization of resources
regarding radiation protection with other organizations including the ICRP.
The ICRP is similar in scope to the NCRP; however, its international membership
brings to bear a variety of perspectives on radiation health issues. The dose limits
of the countries in the European Union are based upon its recommendations. The
NCRP and ICRP have published over 300 monographs containing recommendations
on a wide variety of radiation health issues that serve as the reference documents
from which many regulations are crafted.
NRC Standards for Protection against Radiation

As mentioned above, the NRC has established Standards for Protection against
Radiation (10 CFR 20) to protect radiation workers and the public. The regulations
incorporate a twofold system of dose limitation: (1) the doses to individuals shall not
exceed limits established by the NRC, and (2) all exposures shall be kept ALARA,
social and economic factors being taken into account. The regulations adopt a system
894
recommended by the ICRP, which permits internal doses (from ingested or inhaled
radionuclides) and doses from radiation sources outside the body to be summed,
with a set of limits for the sum.
Summing Internal and External Doses

There are significant differences between external and internal exposures. The dose
from an internal exposure continues after the period of ingestion or inhalation, until
the radioactivity is eliminated by radioactive decay or biologic removal. The exposure may last only a few minutes, for example, in the case of the radionuclide O-15
(T = 122 s), or may last the lifetime of the individual, as is the case for the ingestion
of long-lived Ra-226. The committed dose equivalent (H50,T) is the dose equivalent to a
tissue or organ over the 50 years following the ingestion or inhalation of radioactivity.
The committed effective dose equivalent (CEDE) is a weighted average of the committed dose equivalents to the various tissues and organs of the body:

CEDE = wTH50,T
[21-6]
The NRC adopted the quantity effective dose equivalent and the tissue weighting factors (wT) from ICRP Publication 26 (1977) and the annual limits on intake (ALI)
of radionuclides by workers (discussed below) from ICRP Publication 30 (1978 to
1988), which predate those in the most current ICRP recommendations (ICRP Publication 103, see Chapter 3: Effective Dose).
To sum the internal and external doses to any individual tissue or organ, the
deep dose equivalent (indicated by a dosimeter worn by the exposed individual) and
the committed dose equivalent to the organ are added. The sum of the external and
internal doses to the entire body, called the total effective dose equivalent (TEDE), is
the sum of the deep dose equivalent and the CEDE.
Occupational Dose Limits

The NRCs radiation dose limits are intended to limit the risks of stochastic effects, such
as cancer and genetic effects, and to prevent deterministic effects, such as cataracts, skin
damage, sterility, and hematologic consequences of bone marrow depletion (deterministic and stochastic effects are discussed in Chapter 20). To limit the risk of stochastic
effects, the sum of the external and internal doses to the entire body, the TEDE, may
not exceed 0.05 Sv (5 rem) in a year. To prevent deterministic effects, the sum of the
external dose and committed dose equivalent to any individual organ except the lens of
the eye may not exceed 0.5 Sv (50 rem) in a year. The dose limit to the lens of the eye is
0.15 Sv (15 rem) in a year (Fig. 21-21). As discussed in Chapter 20, the ICRP recently
reviewed the scientific evidence regarding the risk of radiation-induced cataract and
proposed a much more conservative occupational equivalent dose limit for the lens of
the eye (20 mSv per year averaged over 5 years, with no single year dose >50 mSv).
The dose to the fetus of a declared pregnant radiation worker may not exceed
5 mSv (0.5 rem) over the 9-month gestational period and should not substantially
exceed 500 Sv (50 mrem) in any 1 month. Table 21-10 lists the most important
ICRP and NRC radiation dose limits.
Annual Limits on Intake and Derived Air Concentrations

In practice, the committed dose equivalent and CEDE are seldom used in protecting workers from ingesting or inhaling radioactivity. Instead, the NRC has established
annual limits on intake (ALIs), which limit the inhalation or ingestion of radioactive

Annual Dose limit for
Occupationally
Exposed Adults
Total Effective Dose Equivalent
(TEDE) 0.05 Sv
895
Eye Lens
0.15 Sv
Skin
0.5 Sv
Elbow
to Hand
0.5 Sv
Internal
Organs
0.5 Sv
Knees
to Feet
0.5 Sv
Annual Occupational Dose Limits for Adults

NRC 10CFR20
FIGURE 21-21 NRCs occupational radiation dose limits.
material to activities that will not cause radiation workers to exceed any of the limits in
Table 21-10. The ALIs are calculated for the standard person and expressed in units
of microcuries. ALIs are provided for the inhalation and oral ingestion pathways for a
wide variety of radionuclides, some examples of which, relevant to nuclear medicine,
are shown in Table 21-11.
Exposure to airborne activity is also regulated via the derived air concentration
(DAC), which is that concentration of a radionuclide that, if breathed under conditions of light activity for 2,000 h (the average number of hours worked in a year),
will result in the ALI; DACs are expressed in units of Ci/mL. Table 21-11 lists DACs
and ALIs for volatile radionuclides of interest in medical imaging. In circumstances
in which either the external exposure or the internal deposition does not exceed
10% of its respective limits, summation of the internal and external doses is not
required to demonstrate compliance with the dose limits; most nuclear medicine
departments will be able to take advantage of this exemption.
As Low as Reasonably Achievable Principle

Dose limits to workers and the public are regarded as upper limits rather than as
acceptable doses or thresholds of safety. In fact, the majority of occupational exposures in medicine and industry result in doses far below these limits. In addition to
the dose limits specified in the regulations, all licensees are required to employ good
health physics practices and implement radiation safety programs to ensure that
radiation exposures to workers and members of the public are kept ALARA, taking
societal and economic factors into consideration. The ALARA doctrine is the driving
force for many of the policies, procedures, and practices in radiation laboratories and
represents a commitment by both employee and employer to minimize radiation
exposure to staff and the public. For example, the L-shield in the nuclear pharmacy
896
TABLE 21-10NUCLEAR REGULATORY COMMISSION REGULATORY
REQUIREMENTS (NRC 2011) AND INTERNATIONAL

COMMISSION ON RADIOLOGICAL PROTECTION
RECOMMENDATIONS (ICRP 2007) FOR MAXIMUM
PERMISSIBLE DOSE
MAXIMUM PERMISSIBLE ANNUAL DOSE LIMITS
LIMITS
ICRP 2007 mSv
NRC mSv (rem)
Occupational Limits
Effective dose (total effective dose

equivalent)
20/y, averaged over 5 y;

50 in any 1 y
50 (5)
Equivalent dose (Dose equivalent) to

any individual organ (except lens of
eye) including skin; hands, feet
500
500 (50)
Equivalent dose (Dose equivalent) to

the lens of the eye
20/y, averaged over 5 y;

50 in any 1 y
150 (15)
Minor (,18 years old)
Public limits
10% of adult limits
Equivalent dose (Dose equivalent) to an

embryo or fetusb
1 in 9 mo
5 (0.5) in 9 mo
Non-occupational (public) Limits
1.0 (0.1)

equivalent)
1 (0.1)

equivalent) to medical caregivers (e.g.,
at home with I-131 therapy patient)
5 per episode;
20/y maximum
5 (0.5)
Dose equivalent in an unrestricted area
NA
0.02 (0.002) in any

1 hc
These limits are exclusive of natural background and any dose the individual receives for medical purposes;
inclusive of internal committed dose equivalent and external effective dose equivalent (i.e., total effective
dose equivalent). As explained in the text, ICRP and NRC use different (although similar) dose quantities. In
this column the first quantity is the ICRP quantity and the NRC quantity is in parentheses.
b
Applies only to a conceptus of a worker who declares her pregnancy. If the limit exceeds 4.5 mSv (450 mrem)
at declaration, conceptus dose for remainder of gestation is not to exceed 0.5 mSv (50 mrem).
c
This means the dose to an unrestricted area (irrespective of occupancy) shall not exceed 0.02 mSv (2 mrem) in
any 1 h. This is not a restriction of instantaneous dose rate to 0.02 mSv/h (2 mrem/h).
ICRP. The 2007 Recommendations of the International Commission on Radiological Protection. ICRP
Publication 103. Ann ICRP 2007;37:1332.
a
TABLE 21-11EXAMPLES OF DACs AND ALI FOR
OCCUPATIONALLY EXPOSED WORKERS

(NRC REGULATIONS: 10 CFR 20)
RADIONUCLIDE
DAC (mCi/mL)a
ALI (mCi)a INGESTION
I-125
28
3 3 10
40
I-131
2 3 1028
30
Tc-99m
6 310
8 3 104
Xe-133 (gas)
1 3 10
25
24
N/A
Multiply mCi/mL (or mCi) by 37 to obtain kBq/mL (or kBq).

N/A, not applicable.
897
may not be specifically required by a regulatory agency; however, its use has become
common practice in nuclear medicine departments and represents a community
standard against which the departments practices will be evaluated.
The ALARA principle is equivalent to the ICRPs principle of optimization of
protection, stated in Section 21.4. While the NRCs regulations only apply the ALARA
principle to the radiation doses of workers and members of the public, the ICRP
and many other advisory bodies, professional societies, and governmental agencies
recommend the application of the ALARA principle to the radiation doses received
by patients in diagnostic and interventional imaging.
21.9 Prevention of Errors

A goal of radiation protection programs is to reduce the likelihood and severity of accidents involving radiation and radioactive materials. Although this pertains to the protection of staff, patients, and members of the public, this section focuses on protection of
the patient. However, the principles discussed apply to all of these categories of people.
Error prevention is certainly not a field exclusive to medical radiation protection or
even medicine in general. Most of the principles have been developed in other industries such as aviation and nuclear power. The report, To Err Is Human, by the Institute
of Medicine in 2000 (Kohn et al., 2004) focused attention on errors in the delivery
of care to patients. There has been national media attention regarding the delivery of
excessive doses to patients by CT (Bogdanich, 2009) and the professional literature has
reported the delivery of unnecessarily high doses to children and infants from CT due to
the use of adult technique factors (Patterson et al., 2001). Methods for the prevention of
errors in the delivery of care to patients are only briefly discussed in this section.
Efforts in error prevention can be prospective, to avoid errors in the future, or
retrospective, in response to an error or near miss that has occurred, with the goal of
preventing a recurrence. The efforts directed toward prevention of a particular error
should be in proportion to the severity of the consequences of the error and of the likelihood of the error. Methods for the prevention of errors include the development of written procedures and training of staff initially, periodically thereafter, and in response to
incidents or changes in procedures. Policies and procedures may incorporate the use of a
time out in which the staff about to perform a clinical procedure jointly verify that the
intended procedure is to be properly performed on the intended patient; the use of written or computerized checklists to ensure that essential actions are not omitted; checks
of critical tasks by a second person; requiring that certain critical communications be
in writing; employing readback in verbal communications, whereby the person hearing
information repeats it aloud; and measures to avoid distractions and interruptions during critical tasks, such as prohibiting unnecessary conversations unrelated to a task. In
developing procedures, it is advisable to consider what errors might occur, procedures
developed at other institutions, and errors that have occurred in the past. The incorporation of engineered safety features, such as interlocks that prevent unsafe actions,
and automated checks, such as the dose notifications and alerts provided by newer CT
scanners, help avoid accidents due to human errors. The Joint Commission (TJC) has
included similar recommendations in a universal protocol to prevent error prior to surgery, which is one of a number of its national patient safety goals (TJC, 2010).
When an error or a near miss occurs, an analysis should be performed to determine the causes and to devise actions to reduce the likelihood of a recurrence and
possibly, the severity of a recurrence. These analyses attempt to identify the most
basic correctable causes, called root causes. Methods for such root cause analyses
898
may be found in several texts. The actions implemented to avoid a recurrence should
be designed with the root causes in mind.
Safety culture is a set of attributes of an organization, such as a radiology or nuclear
medicine department, that are believed to reduce the likelihood of errors. There are
varying, but similar definitions or descriptions by various organizations focusing on
safety. Safety culture may be defined as the behavior and practices resulting from a
commitment by leaders and individuals to emphasize safety over competing goals to
protect patients, staff, and the public. The following list of traits of a safety culture
was adapted from an NRC policy statement (NRC, 2011):
Leaders
demonstrate a commitment to safety in their decisions and behaviors.

individuals take personal responsibility for safety.
Issues potentially impacting safety are promptly identified, evaluated, and corrected commensurate with their significance.
Work activities are planned and conducted so that safety is maintained.
Opportunities to learn about ways to ensure safety are sought out and implemented.
A work environment is maintained in which people feel free to raise safety concerns.
Communications maintain a focus on safety.
Trust and respect permeate the organization.
Workers avoid complacency and continuously challenge, in a spirit of cooperation,
conditions and activities in order to identify discrepancies that might result in error.
All
The Joint Commission

The Joint Commission (TJC), formerly the Joint Commission on Accreditation of
Healthcare Organizations, is a not-for-profit organization that accredits thousands of
hospitals and other health care facilities in the United States. Most state governments
have conditioned receipt of Medicaid reimbursement on accreditation by TJC or a
similar organization.
TJC defines a sentinel event as an unexpected occurrence involving death or serious physical or psychological injury, or the risk thereof. A sentinel event does not
necessarily indicate that a medical error has occurred. However, each sentinel event
should be promptly investigated and action taken if warranted.
TJC defines as a sentinel event prolonged fluoroscopy with a cumulative dose
exceeding 15 Gy to a single field, a single field being a location on the skin onto
which the x-ray beam is directed. The issue here is the magnitude of the dose
to that portion of the skin that receives the maximal dose. The maximal dose
may result from using several different x-ray beam projections whose beam areas
on the patients skin overlap in a specific location to produce a region of highest radiation dose. It may also result from two or more procedures days or even
months apart.
TJC issues sentinel event alerts (SEAs) which identify serious unanticipated incidents in a health care setting that could or did harm patients. TJC expects accredited facilities to consider information in an SEA when evaluating similar processes
and consider implementing relevant suggestions contained in the SEA or reasonable
alternatives. TJC has issued a SEA, Radiation Risks of Diagnostic Imaging (TJC, 2011).
It warns that health care organizations must seek new ways to reduce exposure to
repeated doses of harmful radiation from these diagnostic procedures. The SEA urges
greater attention to the risk of long-term damage and cumulative harm that can occur
if a patient is given repeated doses of diagnostic radiation. TJC suggests that health
care organizations can reduce risks due to avoidable diagnostic radiation by raising
awareness among staff and patients of the increased risks associated with cumulative
899
doses and by providing the right test and the right dose through effective processes,
safe technology and a culture of safety. The actions suggested include
Use
of imaging techniques other than CT, such as ultrasound or MRI, and collaboration between radiologists and referring physicians about the appropriate use of
diagnostic imaging.
Adherence to the ALARA principle, as well as guidelines from the Society for
Pediatric Radiology, American College of Radiology and the Radiological Society of
North America, for imaging for children and adults.
Assurance by radiologists that the proper dosing protocol is in place for the
patient being treated and review of all dosing protocols either annually or every
2 years.
Expansion of the RSOs role to explicitly include patient safety as it relates to radiation and dosing, as well as education on proper dosing and equipment usage for all
physicians and technologists who prescribe diagnostic radiation or use diagnostic
radiation equipment.
Use of a diagnostic medical physicist in designing and altering CT scan protocols;
centralized quality and safety performance monitoring of all diagnostic imaging
equipment that may emit high amounts of cumulative radiation; testing imaging
equipment initially and annually or every 2 years thereafter; and designing a program for quality control, testing (including daily functional tests) and preventive
maintenance activities.
Investing in technologies that optimize or reduce dose.
21.10 Management of Radiation Safety Programs

The goals of a radiation safety program were discussed in the introductory paragraph
of this chapter. They include maintaining the safety of staff, patients, and members of
the public; compliance with regulations of governmental agencies; and preparedness
for emergencies involving radiation and/or radioactive material. In particular, the
radiation doses of staff and members of the public must not exceed regulatory limits
and should conform to the ALARA principle (equivalent to the ICRPs principle of
optimization), as described earlier in this chapter. Although there are no regulatory
limits to the radiation doses that patients may receive as part of their care, the doses
to patients should be optimized. The methods for achieving these goals are described
in the other sections of this chapter.
A license from a state regulatory agency, or the US NRC, is required to possess
and use radioactive materials in medicine. The license lists the radioactive materials
that may be used, the amounts that may be possessed, and the uses that are allowed.
The licenses for most institutions list the physicians permitted to use and supervise
the use of radioactive material. Licenses also contain requirements, in addition to the
regulations of the state regulatory agency or NRC, regarding procedures and recordkeeping requirements for radioactive material receipt, transportation, use, and disposal. Radioactive material use regulations require cradle-to-grave control of all
radiation sources.
Each institution must designate a person, called the RSO, who is responsible
for the day-to-day oversight of the radiation safety program and is named on the
institutions radioactive material license. At a smaller institution, the RSO may be a
physician or a consultant. A larger institution may have a person whose main duty
is to be the RSO. At very large institutions, the RSO is assisted by a health physics
department.
900
An institution with complex uses of radiation and radioactive material will have
a radiation safety committee that oversees the radiation safety program. A radiation
safety committee typically meets quarterly and approves policies and procedures,
periodically reviews the program, and reviews corrective actions after adverse incidents involving radiation or radioactive material. The radiation safety committee is
comprised of the RSO, representatives from departments with substantial radiation
and radioactive material use (e.g., nuclear medicine, radiology, cardiology, radiation oncology, and nursing), and a member representing hospital management.
A radiation safety program should include an audit program to ensure compliance
with regulatory requirements, accepted radiation safety practices, and institutional
procedures. The audit program should evaluate peoples knowledge and performance
of radiation safetyrelated activities in addition to reviewing records. This is best
accomplished by observing people while they perform tasks related to radiation safety.
Personnel dosimetry reports should be reviewed when they arrive from the dosimetry
vendor and larger or unusual doses should be investigated. The radiation safety program should ensure the training of staff in regulatory requirements, good radiation
safety practices, and institutional procedures; training should be provided initially,
periodically thereafter, and as needed, for example, following incidents, changes in
regulations, or changes in uses of radiation or radioactive material. A review should be
performed of the radiation safety program at least annually, with the findings presented
to the radiation safety committee and executive management. The radiation safety program should be risk informed, that is, the efforts devoted to radiation safety regarding a particular use should be in proportion to the risks associated with that use.
A diagnostic medical physicist is essential to the radiation safety program of a
medical facility performing radiological imaging. A diagnostic medical physicist has
expertise in the diagnostic and interventional applications of x-rays, gamma rays,
ultrasonic radiation, radiofrequency radiation, and magnetic fields; the equipment
associated with their production, use, measurement, and evaluation; the doses of
radiation to patients from these sources; the quality of images resulting from their
production and use; and medical health physics associated with their production
and use. A duty of the diagnostic medical physicist is testing imaging equipment initially, annually thereafter, and after repairs or modifications that may affect the radiation doses to patients or the quality of the images. The diagnostic medical physicist
should also participate in the creation and evaluation of the organizations technical
quality assurance programs, safety programs (e.g., radiation safety and MRI safety)
regarding diagnostic and interventional imaging, and optimization of clinical imaging
protocols. Diagnostic medical physicists commonly design the structural shielding
for rooms containing radiation-producing equipment and perform shielding evaluations, based upon measurements of transmitted radiation, after the installation of the
shielding. Another task occasionally performed is the estimation of the radiation dose
to the embryo or fetus after a pregnant patient received an examination before her
pregnancy was known or when a procedure on a pregnant patient is being considered. The RSO may be a diagnostic medical physicist.
Another task of a radiation safety program is to respond to adverse incidents or
regulatory violations. When such an incident or violation is identified, an analysis of
the causes should be performed and actions to prevent a recurrence should be implemented. In some cases, such as a medical event involving a patient, serious injury or
death of a patient involving medical equipment, or a person receiving a radiation dose
exceeding a regulatory limit, notification of the pertinent state or federal regulatory
agency may be required. When such an incident occurs, action to protect the patient,
staff, or other people should take priority over meeting regulatory r equirements.
901
21.11 IMAGING OF PREGNANT AND POTENTIALLY
PREGNANT PATIENTS
The risks of ionizing radiation to the embryo and fetus must be considered when
imaging female patients of childbearing age using ionizing radiation and performing
therapies using radiopharmaceuticals. These risks vary with the radiation dose and the
stage of gestation and have been discussed in Chapter 20. Each department performing imaging or treatments with ionizing radiation should have policies and procedures
regarding this issue.
In general, the pregnancy status of a patient of childbearing age should be determined prior to an imaging examination or treatment involving ionizing radiation.
Pregnancy status should be determined for female patients from about 12 to 50 years
of age. The measures to determine whether a patient is pregnant depend upon the
radiation dose that an embryo or fetus would receive. For most examinations, a technologist asks the patient whether she could possibly be pregnant, inquiring about the
last menstruation, and documents this information on a form. For a patient who is
a minor, it is best to obtain this information in a private setting without the parents
or a guardian present. For some procedures that would or might impart doses to
an embryo or fetus exceeding 100 mSv, such as possibly prolonged fluoroscopically
guided procedures and multi-phase diagnostic CT examinations of the abdomen or
pelvis, a pregnancy test should also be obtained within 72 hours before the examination or procedure, unless the examination is medically urgent or the possibility of
pregnancy can be eliminated by factors such as a documented hysterectomy. Pregnancy tests should always be obtained before radiopharmaceutical therapies. Pregnancy tests can produce false-negative results and should not preclude questioning the
patient regarding the possibility of pregnancy. There are some examinations that can
be performed without regard to pregnancy due to the very low dose that an embryo
or fetus would receive if the patient were pregnant. These include mammograms and
diagnostic x-ray imaging of the head, arms and hands, and lower legs and feet.
Occasionally imaging is considered for a patient who is known to be pregnant
or suspected of being pregnant. In this case, the radiation dose and potential risks
to the embryo or fetus should be estimated and presented to the imaging physician. Consideration should be given to whether the examination or procedure can be
delayed until after the pregnancy, alternative examinations or procedures that would
not involve ionizing radiation, or that would deliver less radiation to the embryo or
fetus, and ways to modify the examination to reduce the radiation dose to the embryo
or fetus. However, care should be taken to ensure that modifications to reduce the
radiation dose do not cause an inadequate examination that must be repeated. The
imaging physician must obtain informed consent from the patient.
Occasionally it is discovered, after an examination or treatment with ionizing
radiation was performed, that the patient was pregnant at the time of the examination
or treatment. In this case, fetal age, the radiation dose to the embryo or fetus, and the
potential effects on the embryo or fetus should be estimated. The dose estimation is
commonly performed by a diagnostic medical physicist. The patient is then counseled by a physician. In most cases, the risks to the embryo or fetus are small in comparison with the other risks of gestation. Professional consensus is that in no cases
should medical intervention with a pregnancy be considered for estimated doses to
the embryo or fetus that do not exceed 100 mSv; if the dose exceeds 100 mSv, decisions would be guided by the fetal age and individual circumstances. All x-ray imaging examinations, in which the embryo or fetus is not close to the area being imaged,
902
impart doses much less that 100 mSv to the embryo or fetus and most x-ray examinations of the abdomen and pelvis, including single phase diagnostic CT examinations,
impart doses less than 100 mSv. More detailed guidance regarding the pregnant or
possibly pregnant patient is available in ACR, 2008; Wagner et al., 1997.
21.12 Medical Emergencies Involving Ionizing Radiation

Medical institutions should be prepared to respond to emergencies involving radiation
and radioactive materials. These may include incidents within the institution, such as
spills of radioactive material, contamination of personnel with radioactive material, and
medical events involving errors in the administration of radioactive material to patients;
these also include external incidents, such as transportation accidents involving shipments of radioactive material, accidents at nuclear facilities, and radiological or nuclear
terrorism, which could result in contaminated and injured people arriving at medical
facilities. Serious accidents involving substantial radiation exposure or contamination
are rare. Hence, most health care providers are unfamiliar with (and may have unjustified fear of) treating patients who have been exposed to radiation or contaminated with
radioactive material. Radiologists and other health care professionals familiar with the
properties of radiation, its detection, and biological effects (e.g., radiation oncologists,
radiation therapists, health and medical physicists, and nuclear medicine and radiology technologists) may be called upon to assist in such emergencies and should be
prepared to provide guidance and technical assistance to medical staff managing such
patients; they should also assist the medical institutions emergency planner in preparing for such emergencies. The overarching goal in these incidents is to ensure that
critically ill or injured patients are not denied effective medical care due to unfounded
concerns of medical personnel about risk from providing care to the patients.
Potential Sources of Radiation Exposure

and Contamination
Radiation Accidents
Incidents have occurred in which people have been injured by inadvertently coming
into contact with objects or material that they did not realize contained large quantities of radioactive material. Small metal capsules containing high-intensity gammaemitting radioactive sources are commonly used for industrial radiography. Failure
to realize or report the loss of such sources has resulted in a number of severe radiation injuries to individuals who found the sources but did not recognize the danger.
In two cases, stolen radioactive sources once used for cancer therapy were opened,
causing widespread contamination and injury to people (Juarez Mexico in 1983 and
1984, Goinia Brazil in 1987).
A nuclear reactor accident, in which an accidental prompt-critical power excursion caused a steam explosion (Chernobyl in the Ukraine in 1986), killed workers
and emergency responders and released massive amounts of fission products to the
environment. In 2011, a tsunami following a massive earthquake off the coast of
Japan caused the loss of core cooling of multiple reactors at the Fukushima Daiichi
nuclear power facility. Even though the reactors shut down during the earthquake
(stopping further fissioning of the fuel in the reactors), there was still a tremendous
amount of heat being produced by the decay of the fission products already created.
The inability to adequately remove this heat ultimately led to partial core meltdowns.
903
The release of large quantities of fission products from the damaged reactors resulted
in widespread environmental contamination requiring the evacuation of large populations living within 30 to 50 km from the site.
Radiological and Nuclear Terrorism

There has been increasing concern that radioactive material or an improvised nuclear
device (an improvised nuclear weapon, less likely but potentially much more devastating) could be used as a weapon of terrorism. A hidden radioactive source, referred
to as a radiation exposure device, could be used to expose people to large doses of
radiation. Victims of such a device might exhibit the signs and symptoms of the acute
radiation syndrome and possibly also localized radiation skin injuries. Alternatively, a
large activity of radioactive material could be dispersed in a populated area. A device
to do this is referred to as a radiological dispersal device (RDD). An RDD that uses a
chemical explosion to accomplish the dispersion is called a dirty bomb. An RDD,
to be effective, would require a large activity of a radioactive material with a relatively long half-life and that emits abundant penetrating radiation and/or has a high
radiotoxicity if ingested or inhaled. Radionuclides having these qualities and which
are commonly used in very large activities are Co-60 (T1/2 = 5.27 years, 1.17 and
1.33 MeV gamma rays), Sr-90 (T1/2 = 28.8 years, no gamma rays, very highenergy beta
rays), Cs-137 (T1/2 = 30.1 years, 662 keV gamma rays), Ir-192 (T1/2 = 73.8 days, 296 to
612 keV gamma rays), and Am-241 (T1/2 = 433 years, 59.5 keV gamma rays, alpha particles). A dirty bomb detonation could result in a few or dozens of casualties with trauma
and radioactive contamination and a much larger number of uninjured people with
radioactive contamination. Attack on or sabotage of a nuclear power facility or shipments
of large quantities of radioactive material could also cause widespread contamination;
attacks of this nature have the potential to release a much larger quantity of radioactive
material. In recent years, many countries have greatly increased the security of nuclear
facilities and of sources containing large activities of radioactive material (including those
in hospitals and research facilities) to prevent or discourage radiological terrorism.
The detonation of an improvised or stolen nuclear weapon in a city is much less
likely than other forms of terrorism because construction of such a device requires
several kilograms of plutonium or highly enriched uranium and considerable technical expertise, and access to nuclear weapons is tightly controlled. However, with the
proliferation of nuclear weapons and nuclear weapons technology, such an attack
may become more likely in the future. The successful use of such a device in a major
metropolitan area would be devastating, with an area around the weapon demolished
by blast and prompt thermal radiation. If the weapon were detonated at ground level
or a low altitude, there would be extensive radioactive fallout in a pattern determined
by surface and higher-altitude winds. A large number of casualties would be created
rapidly by the blast wave, collapses of structures, injuries from shrapnel and falling
objects, prompt thermal radiation, fires caused by the explosion and prompt radiation consisting of x- and gamma rays, and neutrons from the detonation. Delayed
radiation injuries, particularly the acute radiation syndrome from exposure to gamma
rays from radioactive fallout and cutaneous injuries from beta radiation from fallout
on the skin, would also occur to some people in the fallout zone. The mass casualties,
widespread infrastructure failure, and initial inability to operate in the fallout zone
would greatly impede the response to the incident.
Each medical facility has or should have a plan for responding to radiation
emergencies of varying types and scopes. In particular, the emergency department
(ED) should be expected to receive and treat injured patients with radioactive contamination. Radioactive contamination has only very rarely been an immediate health
904

FIGURE 21-22 Standard protective clothing commonly worn by
ED staff and periodic surveys with
a GM survey meter protect staff
from contamination with radioactive material.
threat to patients and almost never to medical staff. Treatment of severe injuries
or life-threatening medical conditions should take priority over decontamination,
although very simple and quick measures, such as removal of the patients clothing
and wiping exposed skin with mild soap, tepid water, and a damp cloth, will likely
remove most of the contamination. Standard protective clothing commonly worn
by ED staff is sufficient to protect them from radioactive contamination. Periodic
surveys with a GM detector and methods similar to common infection prevention techniques will prevent significant contamination with radioactive material
(Fig. 21-22).
Concern for radiation exposure or radioactive material contamination should
not prevent the use of any of a hospitals diagnostic and treatment capabilities
including the use of medical imaging facilities and operating rooms. Most contamination can be prevented and any contamination that does occur can be managed
and would not render any of these facilities inoperable or unsuitable for other critically ill patients.
Following a large-scale emergency involving radiation and/or radioactive material, it is possible that many people, perhaps some with radioactive contamination,
but without serious injuries or medical conditions requiring emergency treatment,
will arrive at medical centers seeking care. The medical centers plans should include
measures to prevent the emergency departments from becoming overwhelmed by
these people. Each community should prepare to establish decontamination centers away from hospitals. However, each medical center should prepare to promptly
establish a triage, monitoring, and decontamination center away from the emergency
department and to divert to this center people not requiring emergency care. More
detailed information on developing a radiological emergency response plan, staff
training, and sources of expert consultation are widely available (ACR, 2006; Bushberg et al., 2007; HPS, 2011; NCRP, 2001, 2005, 2009, 2010b; REAC/TS, 2011;
REMM, 2011).
Decontamination of People
External radioactive contamination is contamination of a persons clothing, skin,
and perhaps other parts of the body surface with radioactive material. An incident
involving external radioactive contamination may be as simple as a single nuclear
medicine or research laboratory worker who has spilled radioactive material on a
905
small part of his or her body or it could involve dozens or hundreds of people, some
injured and many contaminated, arriving at a medical center after a radiological terrorist incident.
Decontamination is commonly performed by removing contaminated clothing
and washing the contaminated area with soap and water, taking care not to spread
contamination to other parts of the body or onto other people and to ensure that
people do not inhale or ingest the radioactive material. If the radioactive material is
not water soluble, another solvent may be needed. Uncontaminated wounds should
be protected by waterproof dressings during decontamination. Thermal burns of
the skin should be decontaminated by gentle rinsing with sterile saline solution,
but should not be scrubbed to avoid further injury. The progress of decontamination is assessed by surveys using a portable radiation survey meter. Decontamination
is performed until the amount of remaining radioactive contamination approaches
background levels or until decontamination efforts cease to reduce the amount of
remaining contamination. If large numbers of people are contaminated, a greater
amount of residual contamination may have to be accepted. Decontamination efforts
should not abrade the skin to avoid the absorption of radioactive material through
the injured skin.
Contaminated wounds are decontaminated by flushing with sterile saline and
mild scrubbing. Excision of viable tissue to remove radioactive contamination should
be performed only after consulting with an expert.
Personnel with external radioactive contamination, particularly on the upper
parts of their bodies, may also have internal contamination. Radioactive contamination of a person should never delay critical medical care.
Treatment for Internal Radioactive Contamination

The term internal contamination refers to radioactive material in the body. It may
enter the body by inhalation, ingestion, through a wound, by injection, or, for some
radioactive material, by absorption through the skin. It may result from an accident,
poor contamination control in the workplace, a terrorist act or an act of war, or an
administration error in nuclear medicine.
Effective treatment of internal contamination varies with the radionuclide, its
chemical and physical form, and, in some cases, its route into the body. In general,
treatment is more effective the sooner it is begun after the radioactive material is
introduced into the body. Hence, it may be best to begin treatment before definitive
information is available. NCRP Report No. 161, Management of Persons Contaminated
with Radionuclides, provides detailed guidance on the treatment of internal contamination (NCRP, 2008).
An accident in nuclear medicine or a research laboratory or a nuclear reactor accident with a release of fission products can cause internal contamination with radioactive iodine. Radioactive iodine is sequestered by the thyroid gland and retained with
a long biological half-life. The most effective treatment is to administer nonradioactive iodine to block thyroidal uptake of the radioactive iodine. This treatment is
most effective the sooner it is begun, becoming much less effective with time after an
intake. Table 21-12 provides guidance for the treatment of internal contamination
with radioactive iodine.
The amount of internal contamination and the effectiveness of treatment may be
assessed by bioassays. Bioassays are usually either measurements of radiation from
the body (e.g., thyroid counts using a thyroid probe or whole-body counts using a
whole-body counter) or assays of excreta, usually urine.
906
TABLE 21-12GUIDANCE FOR KI ADMINISTRATION: THRESHOLD THYROID
RADIOACTIVE EXPOSURES AND RECOMMENDED DOSAGES

OF KI FOR DIFFERENT RISK GROUPS
PREDICTED
NUMBER OR
NUMBER OR
MILLILITERS (mL)
THYROID GLAND KI Dose FRACTION OF
FRACTION OF
OF ORAL SOLUTION,
EXPOSURE (cGy)d (mg)
130 mg TABLETS 65 mg TABLETS 65 mg/mLa
Adults over 40 y 500
130
2 mL
Adults over
10
18 through 40 y
130
2 mL
Pregnant or
Lactating
Women
130
2 mL
Adolescents,
5
12 through 18 yb
65
1 mL
Children over
5
3 y through 12 y
65
1 mL
Children 1 mo
through 3 y
32
Use KI oral
solutionc
0.5 mL
Infants birth
through 1 mo
16
Use KI oral
solutionc
Use KI oral
solutionc
0.25 mL
See the Home Preparation Procedure for Emergency Administration of Potassium Iodide Tablets to Infants
and Small Children at the URL below.
b
Adolescents approaching adult size (150 lbs) should receive the full adult dose (130 mg)
c
Potassium iodide oral solution is supplied in 1 oz (30 mL) bottles with a dropper marked for 1, 0.5, and
0.25 mL dosing. Each mL contains 65 mg potassium iodide.
d
A centigray (cGy) is a hundredth of a gray (Gy) and is equal to one rad.
NAS: National Research Council. Distribution and administration potassium iodide in the event of a nuclear
incident. Washington, DC: National Academies Press, 2004.
http://www.fda.gov/Drugs/EmergencyPreparedness/BioterrorismandDrugPreparedness/ucm072248.htm
a
Treatment of the Acute Radiation Syndrome

The acute radiation syndrome, discussed in Chapter 20, is typically caused by the
irradiation of the entire body or a large portion of the body to a very large dose
of penetrating radiation, such as x-rays, gamma rays, or neutrons. An acute dose
of x- or gamma rays to the bone marrow exceeding 0.5 Gy is usually necessary to
cause the syndrome; an acute dose of about 3.5 gray will kill about half of exposed
individuals without medical care; and an acute dose exceeding 10 gray will kill
nearly all individuals even with medical care. The lethality is increased by skin
burns or trauma. The most significant adverse effects, for individuals receiving
doses that they may survive, are injury to the lining of the gastrointestinal tract and
the red bone marrow, causing a loss of electrolytes, the entry of intestinal flora into
the body, and a depression or even elimination of white blood cells, increasing the
risk of infection.
The dose received by an individual can be crudely estimated by the severity of
the prodromal phase, particularly the time to emesis. It can be more accurately estimated by the kinetics of circulating white blood cells, particularly lymphocytes, and
by a cytogenetic assay of circulating lymphocytes, which requires a blood sample to
be sent to a specialized laboratory. Treatment involves reverse isolation, support of
electrolytes, support of bone marrow recovery by the administration of cytokines,
perhaps the administration of irradiated red blood cells and platelets, prophylactic
907
treatment with antibiotics, and treatment of infections (Waselenko et al., 2004). If

surgery is needed for injuries, it should either be performed in the first 2 days after
the exposure or should be delayed until the hematopoietic system recovers.
SUGGESTED READINGS
American College of Radiology, ACR Practice Guideline for Imaging Pregnant or Potentially Pregnant
Adolescents and Women with Ionizing Radiation, 2008.
Brateman L. The AAPM/RSNA physics tutorial for residents: radiation safety considerations for diagnostic
radiology personnel. RadioGraphics 1999;19:10371055.
Bushberg JT, Leidholdt EM Jr. Radiation protection. In: Sandler MP et al., eds. Diagnostic nuclear medicine.
4th ed. Philadelphia, PA: Lippincott Williams & Wilkins, 2001.
International Commission on Radiological Protection. ICRP Publication 102: Managing patient dose in
multi-detector computed tomography (MDCT) Annals of the ICRP Volume 37/1 2007
International Commission on Radiological Protection. ICRP Publication 105: Radiological Protection
in Medicine Annals of the ICRP Volume 37 Issue 6.
Miller DL, Vano E, Bartel G, et al. Occupational radiation protection in interventional radiology: A joint
guideline of the cardiolvascular and interventional society of Europe and the society of interventional
radiology. Cardiovasc Intervent Radiol 2010;33:230239.
National Council on Radiation Protection and Measurements. Medical x-ray, electron beam, and gamma-ray
protection for energies up to 50 MeV (equipment design, performance, and use). NCRP Report No. 102.
Bethesda, MD: National Council on Radiation Protection and Measurements, 1989.
National Council on Radiation Protection and Measurements. Radiation protection for medical and allied
health personnel. NCRP Report No. 105. Bethesda, MD: National Council on Radiation Protection and
Measurements, 1989.
National Council on Radiation Protection and Measurements. Limitation of exposure to ionizing radiation.
NCRP Report No. 116. Bethesda, MD: National Council on Radiation Protection and Measurements,
1993.
National Council on Radiation Protection and Measurements. Sources and magnitude of occupational and
public exposures from nuclear medicine procedures. NCRP Report No. 124. Bethesda, MD: National
Council on Radiation Protection and Measurements, 1996.
National Council on Radiation Protection and Measurements. Radionuclide exposure of the embryo/fetus. NCRP
Report No. 128. Bethesda, MD: National Council on Radiation Protection and Measurements, 2000.
National Council on Radiation Protection and Measurements. Radiation protection for procedures performed
outside the radiology department. NCRP Report No. 133. Bethesda, MD: National Council on Radiation
Protection and Measurements, 2000.
National Council on Radiation Protection and Measurements. Ionizing Radiation Exposure of the Population
of the United States. NCRP Report No. 160. Bethesda, MD: National Council on Radiation Protection
and Measurements, 2009.
National Council on Radiation Protection and Measurements. Radiation Dose Management for FluoroscopicallyGuided Interventional Medical Procedures. NCRP Report No. 168. Bethesda, MD: National Council on
Radiation Protection and Measurements, 2010.
National Council on Radiation Protection and Measurements. Proceedings of the 43rd Annual NCRP
Meeting (2007). Adv Radiat Protect Med Health Phys 2008;95(5).
Schueler BA, Vrieze TJ, Bjarnason H, et al. An investigation of operator exposure in interventional
radiology. Radiographics 2006;26:15331541.
Siegel JA. Guide for diagnostic nuclear medicine. Reston, VA: Society of Nuclear Medicine, 2002.
Title 10 (Energy) of the Code of Federal Regulations (CFR), Part 20: Standards for Protection Against
Radiation.
Title 10 (Energy) of the Code of Federal Regulations (CFR), Part 35: Medical Use of By-product Material.
Wagner LK, Archer BR. Minimizing risks from fluoroscopic x-rays; bioeffects, instrumentation and examination,
a credentialing program for physicians. 4th ed. Houston, TX: Partners in Radiation Management,
2004.
Wagner LK, Lester RG, Saldana LR. Exposure of the pregnant patient to diagnostic radiations. 2nd ed. Madison,
WI: Medical Physics Publishing, 1997.
REFERENCES
American Association of Physicists in Medicine. AAPM Task Group 108: PET and PET/CT shielding
requirements. Med Phys 2006;33(1).
908

American Association of Physicists in Medicine. AAPM Dose Check Guidelines version 1.0 04/27/2011,
http://www.aapm.org/pubs/CTProtocols/documents/NotificationLevelsStatement_2011-04-27.pdf,
accessed October 1, 2011
American College of Radiology, Disaster Preparedness for Radiology Professionals, Response to Radiological Terrorism:
A Primer for Radiologists, Radiation Oncologists and Medical Physicists, version 3, 2006. www.acr.org
Bogdanich W. Radiation Overdoses Point Up Dangers of CT Scans, New York Times, October 15, 2009.
Brenner DJ, Hall EJ. Computed tomography: an increasing source of radiation exposure. N Engl J Med
2007;357(22):22772284.
Brenner DJ, Elliston CD, Hall EJ, Berdon WE. Estimated Risks of Radiation-Induced Fatal Cancer from
Pediatric CT. AJR 2001; 176:289296.
Bushberg JT, Kroger LA, Hartman MB, et al. Nuclear/radiological terrorism: Emergency department
management of radiation casualties. J Emerg Med 2007;32(1):7185.
Compagnone G, Baleni MC, Pagan L, et al. Comparison of radiation doses to patients undergoing standard
radiographic examinations with conventional screenfilm radiography, computed radiography and
direct digital radiography. Br J Radiol. 2006;79:899904.
Fletcher DW, Miller DL, Balter S, Taylor MA. Comparison of four techniques to estimate radiation dose to skin
during angiographic and interventional radiology procedures. J Vasc Interv Radiol 2002;13:391397.
Guillet B, et al. Technologist radiation exposure in routine clinical practice with 18F-FDG PET. J Nucl Med
Technol 2005;33:175179.
Hausleiter J, Meyer T, Hermann F, et al. Estimated radiation dose associated with cardiac CT angiography.
JAMA 2009;301(5):500507.
HPS. Emergency Department Management of Radiation Casualties, Health Physics 2011. Society http://hps.
org/hsc/responsemed.html
International Commission on Radiological Protection. The 2007 Recommendations of the International
Commission on Radiological Protection. ICRP Publication 103. Ann ICRP 2007;37:1332.
The Joint Commission 2010 Universal Protocol: http://www.jointcommission.org/standards_information/
up.aspx
The Joint Commission. Sentinel event alert, radiation risks of diagnostic imaging, Issue 47, August 24, 2011
Kohn LT, Corrigan JM, Donaldson MS, eds.; Committee on Quality of Health Care in America, Institute
of Medicine. To err is human: Building a safer health system. Washington DC: National Academies Press,
2000. Accessed January 30, 2004.
Miller DL, Balter S, Cole PE, Lu HT, Berenstein A, Albert R, Schueler BA, Georgia JD, Noonan PT, Russell
EJ, Malisch TW, Vogelzang RL, Geisinger M, Cardella JF, St. George J, Miller GL III, Anderson J.
Radiation doses in interventional radiology procedures: the RAD-IR study Part II: Skin dose. J Vasc
Interv Radiol 2003;14:977990.
Murphy PH, Wu Y, Glaze SA. Attenuation properties of lead composite aprons. Radiology 1993;186:269272.
Nair RR, Rajan B, Akiba S, et al. Background radiation and cancer incidence in Kerala, India-Karanagappally
cohort study. Health Phys 2009;96:5566.
NAS: National Research Council. Distribution and administration potassium iodide in the event of a nuclear
incident. Washington, DC: National Academies Press, 2004.
National Council on Radiation Protection and Measurements. Ionizing Radiation Exposure of the Population
of the United States, NCRP Report No. 93 National Council on Radiation Protection and Measurements,
Bethesda, MD, 1987.
National Council on Radiation Protection and Measurements. Use of Personal Monitors to Estimate Effective
Dose Equivalent and Effective Dose to Workers for External Exposure to Low-LET Radiation, NCRP Report
No. 122. National Council on Radiation Protection and Measurements, Bethesda, MD, 1995.
National Council on Radiation Protection and Measurements. Management of terrorist events involving
radioactive material. NCRP Report No. 138 National Council on Radiation Protection and Measurements,
Bethesda, MD, 2001.
National Council on Radiation Protection and Measurements. Structural shielding design for medical x-ray
imaging facilities. NCRP Report No. 147 National Council on Radiation Protection and Measurements,
Bethesda, MD, 2004.
National Council on Radiation Protection and Measurements. Key elements of preparing emergency
responders for nuclear and radiological terrorism. NCRP Commentary No. 19 National Council on
Radiation Protection and Measurements, Bethesda, MD, 2005.
National Council on Radiation Protection and Measurements. Management of persons contaminated with
radionuclides. NCRP Report No. 161 National Council on Radiation Protection and Measurements,
Bethesda, MD, 2008.
National Council on Radiation Protection and Measurements. Ionizing radiation exposure of the population of
the United States. NCRP Report No. 160 National Council on Radiation Protection and Measurements,
Bethesda, MD, 2009.
909
National Council on Radiation Protection and Measurements. Radiation dose management for fluoroscopicallyguided interventional medical procedures. NCRP Report No. 168. National Council on Radiation
Protection and Measurements, Bethesda, MD, 2010a.
National Council on Radiation Protection and Measurements. Responding to a radiological or nuclear
terrorism incident: a guide for decision makers. NCRP Report No. 165. National Council on Radiation
Protection and Measurements, Bethesda, Maryland, 2010b.
Paterson A, Donald P. Frush DP, Donnelly LF. Helical CT of the body: Are settings adjusted for pediatric
patients? Am J Roentgenol 2001;176:297301.
Proceedings of the Thirty-third Annual Meeting of the National Council on Radiation Protection and
Measurements. The effects of pre- and postconception exposure to radiation. Teratology 1999;59(4).
Radiation Emergency Assistance Center/Training Site (REAC/TS), Oak Ridge Institute for Science and
Education (ORISE) positions the U.S. Department of Energy (DOE), 2011; http://orise.orau.gov/
reacts/
Radiation Emergency Medical Management (REMM), Department of Health and Human Services, 2011;
http://www.remm.nlm.gov/
Seibert JA, Shelton DK, Elizabeth H, et al. Computed radiography x-ray exposure trends. Acad Radiol
1996;3:313318.
Stabin MG, Breitz HB. Breast milk excretion of radiopharmaceuticals: Mechanisms, findings, and radiation
dosimetry. J Nucl Med 2000;41(5):863873.
US Nuclear Regulatory Commission, NUREG-1556. Consolidated Guidance about Materials Licenses.
Volume 9, Program-Specific Guidance about Medical Use Licenses. Revision 2, U.S. Nuclear Regulatory
Commission, Washington, DC, January 2008.
US Nuclear Regulatory Commission, Safety Culture Policy Statement, NUREG/BR-0500, June 2011.
Waselenko JK, MacVittie TJ, BlakelyWF, et al. Medical Management of the Acute Radiation Syndrome:
Recommendations of the Strategic National Stockpile Radiation Working Group K. Ann Intern Med
2004;140(12):10371051.
Williams MB, Krupinski EA, Strauss KJ, et al. Digital radiography image quality: image acquisition. J Am
Coll Radiol 2007;4:371388.
Zuguchi M, Chida K, Taura M, Inaba Y, Ebata A, Yamada S. Usefulness of non-lead aprons in
radiation protection for physicians performing interventional procedures. Radiat Prot Dosimetry
2008;131(4):531534.
Section
V
Appendices
A P P EN D I X
Fundamental Principles
of Physics
A.1 Physics Laws, Quantities, and Units
Laws of Physics
Physics is the study of the physical environment around usfrom the smallest
quarks to the galactic dimensions of black holes and quasars. Much of what is
known about physics can be summarized in a set of laws that describe physical
reality. These laws are based on reproducible results from physical observations
that are consistent with theoretical predictions. A well-conceived law of physics
is applicable in a wide range of circumstances. A physical law that states that the
force exerted by gravity on an individual on the surface of the Earth is 680 N is a
poor example, because it is a description of a single situation only. Newtons law
of gravity, however, describes the gravitational forces between any two bodies at
any distance from each other and is an example of a well-conceived, generally
usable law.
Vector Versus Scalar Quantities

For some quantities, such as force, velocity, acceleration, and momentum, direction
is important, in addition to the magnitude. These quantities are called vector quantities. Quantities that do not incorporate direction, such as mass, time, energy, electrical charge, and temperature, are called scalar quantities.
A vector quantity is represented graphically by an arrow whose length is proportional to the magnitude of the vector. A vector quantity is represented by boldface
type, as in the equation F 5 ma , where F and a are vector quantities and m is a
scalar quantity.
In many equations of physics, such as F 5 ma, a vector is multiplied by a
scalar. In vector-scalar multiplication, the magnitude of the resultant vector is
the product of the magnitude of the original vector and the scalar; however, the
direction of the vector is not changed by the multiplication. If the scalar has
units, the multiplication may also change the units of the vector. Force, for example, has different units than acceleration. Vector-scalar multiplication is shown in
Figure A-1.
Two or more vectors may be added. This addition is performed graphically by
placing the tail of one vector against the head of the other, as shown in Figure A-2.
The resultant vector is that reaching from the tail of the first to the head of the
913
914
Section V Appendices
Original
Vector (V)
Vx2
V x 1/2
V x (1)
FIGURE A-1 Vector-scalar multiplication.
s econd. The order in which vectors are added does not affect the result. A special case
occurs when one vector has the same magnitude but opposite direction of the other;
in this case, the two vectors cancel, resulting in no vector.
International System of Units

As science has developed, many disciplines have devised their own specialized units
for measurements. An attempt has been made to establish a single set of units to be
used across all disciplines of science. This set of units, called the Systeme International
(SI), is gradually replacing the traditional units. The SI establishes a set of seven
base unitsthe kilogram, meter, second, ampere, kelvin, candela, and molewhose
magnitudes are carefully described by standards laboratories. Two supplemental
units, the radian and steradian, are used to describe angles in two and three dimensions, respectively. All other units, called derived units, are defined in terms of these
fundamental units. For representing quantities much greater than or much smaller
than an SI unit, the SI unit may be modified by a prefix. Commonly used prefixes are
shown in Table A-1. A more complete list of prefixes is provided in Appendix C along
with physical constants, geometric formulae, conversion factors, and radiologic data
for elements 1 through 100.
FIGURE A-2 Vector addition.

V2
V1 + V2
V1
Appendix A Fundamental Principles of Physics
915
TABLE A-1COMMONLY USED PREFIXES FOR UNITS

PREFIX
MEANING
centi (c)
102
PREFIX
MEANING
milli (m)
103
kilo (k)
1013
micro (m)
106
mega (M)
1016
nano (n)
10
giga (G)
1019
pico (p)
1012
tera (T)
10112
femto (f)
10
peta (P)
10115
15
A.2 Classical Physics
Mass, Length, and Time

Mass is a measure of the resistance a body has to acceleration. It should not be confused with weight, which is the gravitational force exerted on a mass. The SI unit for
mass is the kilogram (kg). The SI unit for length is the meter (m) and for time is the
second (s).
Velocity and Acceleration

Velocity, a vector quantity, is the rate of change of position with respect to time:

v 5 x/t
[A-1]
The magnitude of the velocity, called the speed, is a scalar quantity. The SI unit for
velocity and speed is the meter per second (m/s). Acceleration, also a vector quantity,
is defined as the rate of change of velocity with respect to time:

a 5 v/t
[A-2]
The SI unit for acceleration is the meter per second per second (m/s2).
Forces and Newtons Laws

Force, a vector quantity, is a push or pull. The physicist and mathematician Isaac
Newton proposed three laws regarding force and velocity:
1. Unless acted upon by external force, an object at rest remains at rest and an
object in motion remains in uniform motion. That is, an objects velocity remains
unchanged unless an external force acts upon the object.
2. For every action, there is an equal and opposite reaction. That is, if one object
exerts a force on a second object, the second object also exerts a force on the first
that is equal in magnitude but opposite in direction.
3. A force acting upon an object produces an acceleration in the direction of the
applied force:

F 5 ma
[A-3]
The SI unit for force is the newton (N): 1 N is defined as 1 kg-m/s2. There are
four types of forces: gravitational, electrical, magnetic, and nuclear. These forces are
discussed later.
916
Energy
Kinetic Energy
Kinetic energy, a scalar quantity, is a property of moving matter that is defined by the
following equation:

Ek 5 mv2
[A-4]
where Ek is the kinetic energy, m is the mass of the object, and v is the speed of the
object. (Einstein discovered a more complicated expression that must be used for
objects with speeds approaching the speed of light.) The SI unit for all forms of
energy is the joule (J): 1 J is defined as 1 kg-m2/s2.
Potential Energy
Potential energy is a property of an object in a force field. The force field may be a
gravitational force field. If an object is electrically charged, it may be an electrical
force field caused by nearby charged objects. If an object is very close to an atomic
nucleus, it may be influenced by nuclear forces. The potential energy (Ep) is a function of the position of the object; if the object changes position with respect to the
force field, its potential energy changes.
Conservation of Energy
The total energy of an object is the sum of its kinetic and potential energies:

Etotal 5 Ek 1 Ep
[A-5]
Aside from friction, the total energy of the object does not change. For example,
consider a brick held several feet above the ground. It has a certain amount of gravitational potential energy by virtue of its height, but it has no kinetic energy because
it is at rest. When released, it falls downward with a continuous increase in kinetic
energy. As the brick falls, its position changes and its potential energy decreases.
The sum of its kinetic and potential energies does not change during the fall. One
may think of the situation as one in which potential energy is converted into kinetic
energy.
Momentum
Momentum, like kinetic energy, is a property of moving objects. However, unlike
kinetic energy, momentum is a vector quantity. The momentum of an object is
defined as follows:

p 5 mv
[A-6]
where p is the momentum of an object, m is its mass, and v is its velocity. The
momentum of an object has the same direction as its velocity. The SI unit of momentum is the kilogram-meter per second (kg-m/s). It can be shown from Newtons laws
that, if no external forces act on a collection of objects, the total momentum of the
set of objects does not change. This principle is called the law of conservation of
momentum.
917
A.3 Electricity and Magnetism
Electricity
Electrical Charge
Electrical charge is a property of matter. Matter may have a positive charge, a negative
charge, or no charge. The charge on an object may be determined by observing how
it behaves in relation to other charged objects. The coulomb (C) is the SI derived unit
of electric charge. It is defined in terms of two SI base units as the charge transported
by a steady current of one ampere in one second. The smallest magnitude of charge is
that of the electron. There are approximately 1019 electron charges per coulomb.
A fundamental law of physics states that electrical charge is conserved. This means
that, if charge is neither added to nor removed from a system, the total amount of
charge in the system does not change. The signs of the charges must be considered in
calculating the total amount of charge in a system. If a system contains two charges
of the same magnitude but of opposite signfor example, in a hydrogen atom, in
which a single electron (negative charge) orbits a single proton (positive charge)
the total charge of the system is zero.
Electrical Forces and Fields

The force (F) exerted on a charged particle by a second charged particle is described
by the equation:

F 5 kq1q2/r2
[A-7]
where q1 and q2 are the charges on the two particles, r is the distance between the
two particles, and k is a constant. The direction of the force on each particle is either
toward or away from the other particle; the force is attractive if one charge is negative
and the other positive, and it is repulsive if both charges are of the same sign.
An electrical field exists in the vicinity of electrical charges. To measure the electrical
field strength at a point, a small test charge is placed at that point and the force on the
test charge is measured. The electrical field strength (E) is then calculated as follows:

E 5 F/q
[A-8]
where F is the force exerted on the test charge and q is the magnitude of the test charge.
Electrical field strength is a vector quantity. If the test charge q is positive, E and F have
the same direction; if the test charge is negative, E and F have opposite directions.
An electrical field may be visualized as lines of electrical field strength existing
between electrically charged entities. The strength of the electrical field is depicted as
the density of these lines. Figure A-3 shows the electrical fields surrounding a point
charge and between two parallel plate electrodes.
An electrical dipole has no net charge but does possess regional distribution of
positive and negative charge. When placed in an electrical field, the dipole tends to
align with the field because of the torque exerted on it by that field (attraction by
unlike charges, repulsion by like charges). The electrical dipole structure of certain
natural and man-made crystals is used in ultrasound imaging devices to produce and
detect sound waves, as described in Chapter 16.
Electrical Forces on Charged Particles

When a charged particle is placed in an electrical field, it experiences a force that is
equal to the product of its charge q and the electrical field strength E at that location:
918
+
+
FIGURE A-3 Electrical fields around a point charge (left) and between two parallel charged electrodes
(right).
F 5 qE
If the charge on the particle is positive, the force is in the direction of the electrical
field at that position; if the charge is negative, the force is opposite to the electrical
field. If the particle is not restrained, it experiences an acceleration:
a 5 qE/m

where m is the mass of the particle.
Electrical Current
Electrical current is the flow of electrical charge. Charge may flow through a solid,
such as a wire; through a liquid, such as the acid in an automobile battery; through a
gas, as in a fluorescent light; or through a vacuum, as in an x-ray tube. Current may
be the flow of positive charges, such as positive ions or protons in a cyclotron, or it
may be the flow of negative charges, such as electrons or negative ions. In some cases,
such as an ionization chamber radiation detector, there is a flow of positive charges
in one direction and a flow of negative charges in the opposite direction. Although
current is often the flow of electrons, the direction of the current is defined as the
flow of positive charge (Fig. A-4). The SI unit of current is the ampere (A): 1 A is 1 C
of charge passing a point in an electric circuit per second.
Electrical Potential Difference

Electrical potential, more commonly called voltage, is a scalar quantity. Voltage is the
difference in the electrical potential energy of an electrical charge at two positions,
divided by the charge:

Vab 5 (Epa 2 Epb)/q
[A-9]
where Epa is the electrical potential energy of the charge at location a, Epb is the electrical potential energy of the same charge at location b, and q is the amount of charge.
It is meaningless to specify the voltage at a single point in an electrical circuit; it must
be specified in relation to another point in the circuit. The SI unit of electrical potential is the volt (V): 1 V is defined as 1 J/C, or 1 kg m2/s22C.
Potential is especially useful in determining the final kinetic energy of a charged
particle moving between two electrodes through a vacuum, as in an x-ray tube.
According to the principle of conservation of energy (Equation A-5), the gain in
kinetic energy of the charged particle is equal to its loss of potential energy:

Ek2final 2 Ek2initial 5 Ep2initial 2 Ep2final
[A-10]

Current
e
e
e
919
FIGURE A-4 Direction of electrical current. A. A situation

is shown in which electrons are flowing to the left. B. Positive
charges are shown moving to the right. In both cases, the direction of current is to the right. C. Although the electrons in an
x-ray tube pass from cathode to anode, current technically flows
in the opposite direction.
Current
p+
p+
p+
(Vacuum)
Current
e
e
Anode +
e
Electron Flow
Cathode
(Vacuum)
Assuming that the charged particle starts with no kinetic energy (Ek2initial 5 0) and
using Equations A-9, and A-10 becomes

Ek2final 5 qV
[A-11]
where q is the charge of the particle and V is the potential difference between the
two electrodes.
For example, the final kinetic energy of an electron (charge 5 1.602 3 10219 C)
accelerated through an electrical potential difference of 100 kV is

Ek2final 5 qV (1.602 3 10219 C) (100 kV) = 1.602 3 10214 J
The joule is a rather large unit of energy for subatomic particles. In atomic and
nuclear physics, energies are often expressed in terms of the electron volt (eV). One
electron volt is the kinetic energy developed by an electron accelerated across a
potential difference of 1 V. One electron volt is equal to 1.602 3 10219 J.
For example, the kinetic energy of an electron, initially at rest, that is accelerated
through a potential difference of 100 kV is

Ek 5 qV 5 (1 electron charge) (100 kV) 5 100 keV
One must be careful not to confuse the units of potential difference (V, kV, and
MV) with units of energy (eV, keV, and MeV).
920
Electrical Power
Power is defined as the rate of the conversion of energy from one form to another
with respect to time:

P 5 E/t
[A-12]
For an example, an automobile engine converts chemical energy in gasoline

into kinetic energy of the automobile. The amount of energy converted into kinetic
energy per unit time is the power of the engine. The SI unit of power is the watt (W):
1 W is defined as 1 J/s.
When electrical current flows between two points of different potential, the
potential energy of each charge changes. The change in potential energy per unit
charge is the potential difference (V), and the amount of charge flowing per unit time
is the current (i). Therefore, the electrical power (P) is equal to

P 5 iV
[A-13]
where i is the current and V is the potential difference. Because 1 V 5 1 J/C and 1 A.
5 1 C/s, 1 W 5 1 A2V:

(1 A)(1 V) 5 (1 C/s)(1 J/C) 5 1 J/s 5 1 W
For example, if a potential difference of 12 V applied to a heating coil produces a

current of 1 A, the rate of heat production is

`P 5 iV 5 (1 A) (12 V) 5 12 W
Direct and Alternating Current

Electrical power is normally supplied to equipment as either direct current (DC) or
alternating current (AC). DC power is provided by two wires connecting the equipment to a power source. The power source maintains a constant potential difference
between the two wires (Fig. A-5A). Many electronic circuits require DC, and chemical batteries produce DC.
AC power is usually provided as single-phase or three-phase AC. Single-phase AC
is supplied using two wires from the power source. The power source produces a
potential difference between the two wires that varies sinusoidally with time (see
Fig. A-5B). Single-phase AC is specified by its amplitude (the maximal voltage) and
its frequency (the number of cycles per unit time). Normal household and building power is single-phase 117-V AC. Its actual peak amplitude is 1.4 117 V, or
165V.
AC power is usually supplied to buildings and to heavy-duty electrical equipment as three-phase AC. It uses three or four wires between the power source and the
Voltage
360
120
240
360
time
1 / 60 s
1 / 60 s
FIGURE A-5 A. Direct current. B. Single-phase AC. C. Three-phase AC, comprised of three single-phase AC
sources, each separated by 120 degrees over the repeating 360 degrees cycle.
921
equipment. When present, the fourth wire, called the neutral wire, is maintained at
ground potential (the potential of the earth). The power source provides a sinusoidally varying potential on each of the three wires with respect to the neutral wire.
Voltages on these three wires have the same amplitudes and frequencies; however,
each is a third of a cycle out of phase with respect to the other two, as shown in
FigureA-5C. Each cycle is assigned a total of 360 degrees, and therefore each cycle
in a three-phase circuit is 120 degrees out of phase with the others.
AC is produced by electric generators and used by electric motors. The major
advantage of AC over DC is that its amplitude may be easily increased or decreased
using devices called transformers, as described later. AC is converted to DC for many
electronic circuits.
Ground potential is the electrical potential of the earth. One of the two wires supplying single-phase AC is maintained at ground potential, as is the neutral wire used
in supplying three-phase AC. These wires are maintained at ground potential by connecting them to a metal spike driven deep into the earth.
Conductors, Insulators, and Semiconductors

When a potential difference is applied across a conductor, it causes an electrical
current to flow. In general, increasing the potential difference increases the current.
The quantity resistance is defined as the ratio of the applied voltage to the resultant
current:

R 5 V/i
[A-14]
The SI unit of resistance is the ohm (): 1 is defined as 1 volt per ampere.
Current is defined as the flow of positive charge. In solid matter, however, current
is caused by the movement of negative charges (electrons) only. Based on the amount
of current generated by an applied potential difference, solids may be roughly classified as conductors, semiconductors, or insulators. Metals, especially silver, copper,
and aluminum, have very little resistance and are called conductors. Some materials,
such as glass, plastics, and fused quartz, have very large resistances and are called
insulators. Other materials, such as selenium, silicon, and germanium, have intermediate resistances and are called semiconductors.
The band theory of solids explains the conduction properties of solids. The outershell electrons in solids exist in discrete energy bands. These bands are separated by
gaps; electrons cannot possess energies within the gaps. For an electron to be mobile,
there must be a nearby vacant position in the same energy band into which it can
move. In conductors, the highest energy band occupied by electrons is only partially
filled, so the electrons in it are mobile. In insulators and semiconductors, the highest
occupied band is completely filled and electrons are not readily mobile.
The difference between insulators and semiconductors is the width of the forbidden gap between the highest occupied band and the next higher band; in semiconductors, the gap is about 1 eV, whereas in insulators, it is typically greater than 5 eV.
At room temperature, thermal energy temporarily promotes a small fraction of the
electrons from the valence band into the next higher band. The promoted electrons
are mobile. Their promotion also leaves behind vacancies in the valence band, called
holes, into which other valence band electrons can move. Because the band gap of
semiconductors is much smaller than that of insulators, a much larger number of
electrons exist in the higher band of semiconductors, and the resistance of semiconductors is therefore less than that of insulators. Reducing the temperature of insulators and semiconductors increases their resistance by reducing the thermal energy
available for promoting electrons from the valence band.
922
For certain materials, including most metallic conductors, the resistance is not
greatly affected by the applied voltage. In these materials, the applied voltage and
resultant current are nearly proportional:

V 5 iR
[A-15]
This equation is called Ohms law.
Magnetism
Magnetic Forces and Fields
Magnetic fields can be caused by moving electrical charges. This charge motion may
be over long distances, such as the movement of electrons through a wire, or it may
be restricted to the vicinity of a molecule or atom, such as the motion of an unpaired
electron in a valence shell. The basic features of magnetic fields can be illustrated by
the bar magnet. First, the bar magnet, made of an iron ferromagnetic material, has a
unique atomic electron orbital packing scheme that gives rise to an intense magnetic
field due to the constructive addition of magnetic fields arising from unpaired spinning electrons in different atomic orbital shells. The resultant magnet has two poles
and experiences a force when placed in the vicinity of another bar magnet or external
magnetic field. By convention, the pole that points north under the influence of the
earths magnetic field is termed the north pole, and the other is called the south pole.
Experiments demonstrate that like poles of two magnets repel each other and unlike
poles attract. When a magnet is broken in two, each piece becomes a new magnet,
with a north and a south pole. This phenomenon continues on to the atomic level.
Therefore, the simplest magnetic structure is the magnetic dipole. In contrast, the
simplest electrical structure, the isolated point charge, is unipolar.
Magnetic fields may be visualized as lines of magnetic force. Because magnetic
poles exist in pairs, the magnetic field lines have no beginning or end and, in fact,
circle upon themselves. Figure A-6A shows the magnetic field surrounding a bar
magnet. A current-carrying wire also produces a magnetic field that circles the wire,
as shown in Figure A-6B. Increasing the current flowing through the wire increases
the magnetic field strength. The right-hand rule allows the determination of the magnetic field direction by grasping the wire with the thumb pointed in the direction of
the current (opposite the electron flow). The fingers will then circle the wire in the
direction of the magnetic field. When a current carrying wire is curved in a loop,
the resultant concentric lines of magnetic force overlap and augment the total local
magnetic field strength inside the loop, as shown in Figure A-6C. A coiled wire,
called a solenoid, results in even more augmentation of the magnetic lines of force
within the coil, as shown in Figure A-6D. The field strength depends on the number
of turns in the coil over a fixed distance. An extreme example of a solenoid is found
in most magnetic resonance scanners; in fact, the patient resides inside the solenoid
during the scan. A further enhancement of magnetic field strength can be obtained
by applying a greater current through the wire or by placing a ferromagnetic material
such as iron inside the solenoid. The iron core in this instance confines and augments
the magnetic field. The solenoid is similar to the bar magnet discussed previously;
however, the magnetic field strength may be changed by varying the current. If the
current remains fixed (e.g., DC), the magnetic field also remains fixed; if the current
varies (e.g., AC), the magnetic field varies. Solenoids are also called electromagnets.
Current loops behave as magnetic dipoles. Like the simple bar magnet, they tend to
align with an external magnetic field, as shown in Figure A-7. An electron in an atomic
923
N
e-
S
Current
ee-
FIGURE A-6 Magnetic field descriptions: surrounding a bar magnet (A), around a wire (B), about a wire
loop (C), and about a coiled wire (D).
Unpaired
e-
+ +
++ +
+
Electron
Atom
FIGURE A-7 A spinning charge has an associated magnetic field, such as the electron (left). An unpaired
electron in an atomic orbital produces a magnetic field (right). Each produces magnetic dipoles that will
interact with an external magnetic field.
924
orbital is a current loop about the nucleus. A spinning charge such as the proton
also may be thought of as a small current loop. These current loops act as magnetic
dipoles and tend to align with external magnetic fields, a tendency that gives rise to the
macroscopic magnetic properties of materials (discussed further in Chapter 12.).
The SI unit of magnetic field strength is the tesla (T). An older unit of magnetic
field strength is the gauss (G): 1 T is equal to 104 G. By way of comparison, the earths
magnetic field is approximately 0.05 to 0.1 mT whereas magnetic fields used for
magnetic resonance imaging typically range from 1 T.
Magnetic Forces on Moving Charged Particles

A magnetic field exerts a force on a moving charge, provided that the charge crosses
the lines of magnetic field strength. The magnitude of this force is proportional to (a)
the charge; (b) the speed of the charge; (c) the magnetic field strength, designated B;
and (d) the direction of charge travel with respect to the direction of the magnetic
field. The direction of the force on the moving charge can be determined through
the use of the right-hand rule. It is perpendicular to both the direction of the charges
velocity and the lines of magnetic field strength. Because an electrical current consists
of moving charges, a current-carrying wire in a magnetic field will experience a force
(a torque). This principle is used in devices such as electric motors.
Electromagnetic Induction
In 1831, Michael Faraday discovered that a moving magnet induces an electrical current in a nearby conductor. His observations with magnets and conducting wires led
to the findings listed here. The major ideas are illustrated in Figure A-8.
1. A changing magnetic field induces a voltage (electrical potential difference) in
a nearby conducting wire and causes a current to flow. An identical voltage is
induced whether the wire moves with respect to the magnetic field position or
the magnetic field moves with respect to the wire position. A stationary magnetic
field does not induce a voltage in a stationary wire.
2. A stronger magnetic field results in a stronger induced voltage. The voltage is proportional to the number of magnetic field lines cutting across the wire conductor
per unit time. If the relative speed of the wire or the magnet is increased with
respect to the other, the induced voltage will be larger because an increased number of magnetic field lines will be cutting across the wire conductor per unit time.
3. A 90-degree angle of the wire conductor relative to the magnetic field will provide the greatest number of lines to cross per unit distance, resulting in a higher
induced voltage than for other angles.
4. When a solenoid (wire coil) is placed in a magnetic field, the magnetic field lines
cut by each turn of the coil are additive, causing the resultant induced voltage to
be directly proportional to the number of turns of the coil.
The direction of current flow caused by an induced voltage is described by Lenzs
law (Fig. A-9: An induced current flows in a direction such that its associated magnetic
field opposes the magnetic field that induced it. Lenzs law is an important concept that
describes self-induction and mutual induction.
Self-Induction
A time-varying current in a coil wire produces a magnetic field that varies in the
same manner. By Lenzs law, the varying magnetic field induces a potential difference
across the coil that opposes the source voltage. Therefore, a rising and falling source
voltage (AC) creates an opposing falling and rising induced voltage in the coil. This

e-
Wire
e-
925
Stationary Magnet
e-
e-
e-
in motion
ee-
e-
in motion
eFIGURE A-8 Faraday experiments: changing magnetic fields and induction effects.
A. A stationary magnetic field is depicted, with no electron flow occurring. B. The
magnetic field is moving towards the wire loop, resulting in the induction of an electromotive force that results in electron flow in a clockwise direction. C. The magnetic field is moving away from the position indicated in (B), resulting in the reverse,
counterclockwise flow of electrons in the wire loop.
phenomenon, called self-induction, is used in autotransformers that provide a variable

incremental voltage.
Mutual Induction
A primary wire coil carrying AC produces a time-varying magnetic field. When a secondary wire coil is located under the influence of this magnetic field, a time-varying
potential difference across the secondary coil is similarly induced, as shown in
926
Magnetic field moving towards wire coil

induces an opposing magnetic field
and associated current flow
Induced
magnetic field
Magnet in motion
FIGURE A-9 Lenzs law, demonstrating mutual induction between a moving magnetic field and a coiled
wire conductor.
Figure A-9. The amplitude of the induced voltage, Vs, can be determined from the
following equation, known as the law of transformers:

Vp Vs
N = N
p
s
[A-16]
where Vp is the voltage amplitude applied to the primary coil, Vs is the voltage amplitude induced in the secondary coil, Np is the number of turns in the primary coil,
and Ns is the number of turns in the secondary coil. With a predetermined number
of turns on the primary and secondary coils, this mutual inductance property can
increase or decrease the voltage in electrical circuits. The devices that provide this
change of voltage (current changes in the opposite direction) are called transformers.
Further explanation of their use in x-ray generators is covered in Chapter 6.
Electric Generators and Motors

The electric generator utilizes the principles of electromagnetic induction to convert
mechanical energy into electrical energy. It consists of a coiled wire mounted on a
rotor between the poles of a strong magnet, as shown in Figure A-10A. As an external
mechanical energy source rotates the coil (e.g., hydroelectric turbine generator), the
wires in the coil cut across the magnetic force lines, resulting in a sinusoidally varying
potential difference, the polarity of which is determined by the wire approaching or
receding from one pole of the magnet. The generator serves as a source of AC power.
The electric motor converts electrical energy into mechanical energy. It consists
of a coil of wires mounted on a freely rotating axis (rotor) between the poles of a fixed
magnet, as shown in Figure A-10B. When AC flows through the coil, an increasing
and decreasing magnetic field is generated, the coil acting as a magnetic dipole. This
dipole tends to align with the external magnetic field, causing the rotor to turn. As
the dipole approaches alignment, however, the AC and thus, the magnetic field of the
rotor reverse polarity, causing another half-turn rotation to achieve alignment. The
alternating polarity of the rotors magnetic field causes a continuous rotation of the
coil wire mounted on the rotor as long as AC is applied.
Magnetic Properties of Matter

The magnetic characteristics of materials are determined by atomic and molecular
structures related to the behavior of the associated electrons. Three categories of
927
Rotating current loop
AC
out
Commutator ring
S
Rotation caused by
external force
e.g. water turbine
The rotating current wire loop in the static magnetic field creates an induced current
occuring with an alternating potential difference.
Current
Time
Rotating current loop
AC
in
Commutator ring
Induced rotation of the current loop is caused by the variable magnetic field within current loop
associated with the applied alternating current interacting with the static magnetic field.
FIGURE A-10 Electrical and mechanical energy conversion: electric generators (A) and electric motors (B).
magnetic properties are defined: diamagnetic, paramagnetic, and ferromagnetic.

Their properties arise from the association of moving charges (electrons) and magnetic fields.
Diamagnetic Materials
Individual electrons orbiting in atomic or molecular shells represent a current loop
and give rise to a magnetic field. The various electron orbits do not fall in any preferred plane, so the superposition of the associated magnetic fields results in a net
magnetic field that is too small to be measured. When these atoms or molecules are
placed in a changing magnetic field, however, the electron motions are altered by the
928
induced electrical motive force to form a reverse magnetic field opposing the applied
magnetic field. Diamagnetic materials, therefore, cause a depletion of the applied
magnetic field in the local micromagnetic environment. An example of a diamagnetic
material is calcium, and its magnetic properties are described in Chapter 12.
Paramagnetic Materials
Based on the previous discussion, it would seem likely that all elements and molecules would behave as diamagnetic agents. In fact, all have diamagnetic properties,
but some have additional properties that overwhelm the diamagnetic response. Electrons in atomic or molecular orbitals orient their magnetic fields in such a way as to
cancel in pairs; however, in those materials that have an odd number of electrons, one
electron is unpaired, exact cancellation cannot occur, and a magnetic field equivalent
to one electron orbital results. Depending on orbital structures and electron filling
characteristics, fractional unpaired electron spins can also occur, resulting in variations in the magnetic field strength. When placed in an external magnetic field, the
magnetic field of the material caused by the unpaired electron aligns with the applied
field. Paramagnetic materials, having unpaired electrons, locally augment the micromagnetic environment. The magnetic overall properties o the atom result from these
paramagnetic effects as well as the diamagnetic effects discussed earlier, with the
paramagnetic characteristics predominating. An example of a paramagnetic material
is gadolinium, an element used in magnetic resonance imaging contrast agents, as
described in Chapters 12 and 13.
Ferromagnetic Materials
Iron, nickel, and cobalt can possess intrinsic magnetic fields and will react strongly in
an applied magnetic field. These ferromagnetic materials are transition elements that
have an unorthodox atomic orbital structure: electrons will fill the outer orbital shells
before the inner orbitals are completely filled. The usual spin cancellation of the
electrons does not occur, resulting in an unusually high atomic magnetic moment.
While they are in a random atomic (elemental) or molecular (compound) arrangement, cancellation of the dipoles occurs and no intrinsic magnetic field is manifested; however, when the individual magnetic dipoles are nonrandomly aligned by
an external force (e.g., a strong electrical field), a constructive enhancement of the
individual atomic magnetic moments gives rise to an intrinsic magnetic field that
reacts strongly in an applied magnetic field. Permanent magnets are examples of a
permanent nonrandom arrangement of local magnetic domains of the transition elements. Ferromagnetic characteristics dominate over paramagnetic and diamagnetic
interactions. These materials become dangerous in the vicinity of a strong magnetic
field, and must be carefully identified and avoided in an around magnetic resonance
imaging scanners, as explained in the MR safety section of Chapter 13.
APPENDIX
Digital Computers
Digital computers were originally designed in the 1940s and early 1950s to perform
mathematical computations and other information processing tasks very quickly.
Since then, they have come to be used for many other purposes, including information display, information storage, and, in conjunction with computer networks,
information transfer and communications.
Computers, mechanical, electrical, or part mechanical and part electrical, were
constructed that represented numerical data in analog form. They have been almost
entirely supplanted by digital electronic computers. The explosive growth of such
computers was made possible by the development of the integrated electronic circuit
(chip), which today can have billions of microscopic electronic components such
as transistors and capacitors. In this appendix, the term computer will refer exclusively to programmable digital electronic computers. Analog and digital representation of information was discussed in Chapter 5.
Computers were introduced in medical imaging in the early 1970s and have
become increasingly important since that time. Their first uses in medical imaging
were mostly in nuclear medicine, where they were used to acquire series of images
depicting the organ-specific kinetics of radiopharmaceuticals and to extract physiological information from the images. Today, computers are essential to many imaging
modalities, including x-ray computed tomography (CT), magnetic resonance imaging (MRI), single photon emission computed tomography (SPECT), and positron
emission tomography (PET).
Any function that can be performed by a computer can also be performed by a
hardwired electronic circuit. The advantage of the computer over a hardwired circuit
is its flexibility. The function of the computer can be modified merely by changing
the program that controls the computer, whereas modifying the function of a hardwired circuit usually requires replacing the circuit. Although the computer is a very
complicated and powerful data processing device, the actual operations performed
by a computer are very simple. The power of the computer is mainly due to its speed
in performing these operations and its ability to store large volumes of data.
A workstation is a computer designed for use by a single person at a time. A workstation is usually equipped with one or more display monitors for the visual display
of information, a keyboard for the entry of alphanumeric information, and a pointing
device, such as a mouse.
B.1 Components and Operation of Computers

A computer consists of a central processing unit (CPU), main memory, and input/
output (I/O) devices, connected by data pathways called data buses (Fig. B-1). The
main memory of the computer stores the program (sequence of instructions) being
executed and the data being processed. The CPU executes the instructions in the
program to process the data.
929
930
Mass storage devices
Floppy
disk
Main
memory
CPU
Hard
disk
Memory bus
Bus
interface
Disk
controller
Optical
disk
I/O bus
Analog
data
Acquisition
interface
Interface
Keyboard
Mouse
Network
interface
Video
interface
Video
monitor
Analog
video
signal
To network
FIGURE B-1 Components of a computer.
Digital signals and binary digital signals in particular are discussed in Chapter 5.
Although digital signals need not be binary and some very fast computer networks
use nonbinary digital signals, essentially all computers today store and transfer information in binary form.
A group of wires used to transfer data between several devices is called a data bus.
Each device connected to the bus is identified by an address or a range of addresses.
Only one device at a time can transmit information on a bus, and in most cases, only
one device receives the information. The sending device transmits both the information and the address of the device that is intended to receive the information. Some
buses use separate wires for the data and addresses, whereas others transmit both
over the same wires. The width of a bus refers to the number of wires used to transmit data. For example, a 32-bit bus transmits 32 bits of data simultaneously. Buses
usually have additional wires, such as a ground wire to establish a reference potential
for the voltage signals on the other wires, wires to carry control signals, and wires at
one or more constant voltages to provide electrical power to circuit boards attached
to the bus.
I/O devices enable information to be entered into and retrieved from the computer and to be stored. The I/O devices of a computer usually include a keyboard, a
mouse or other pointing device, a display interface and display monitor, several mass
storage devices, and often a printer. The keyboard, pointing device, display interface, and display monitor enable an operator to communicate with the computer.
Mass storage devices permit the storage of large volumes of programs and data. They
include magnetic disk drives, optical disk drives, and magnetic tape drives.
When the CPU or another device sends data to memory or a mass storage device,
it is said to be writing data, and when it requests data stored in memory or on a storage device, it is said to be reading data. Computer memory and data storage devices
permit either random access or sequential access to the data. The term random access
describes a storage medium in which the locations of the data may be read or written
in any order, and sequential access describes a medium in which data storage locations
can only be accessed in a serial manner. Most solid-state memories, magnetic disks,
and optical disks are random access, whereas magnetic tape typically permits only
sequential access.
Appendix B Digital Computers
931
Main Memory
Main memory provides temporary storage for the instructions of the computer
program being executed and the information being processed by the computer.
Main memory is used for these functions instead of mass storage devices because
the access time is much shorter and the data transfer rate between the CPU and
the main memory is much faster than those between the CPU and the mass storage
devices.
Main memory consists of a large number of data storage locations, each of which
contains the same number of bits (usually 1 byte, which is composed of 8 bits). Each
bit can store only one of two values, represented by the binary digits 0 and 1. Bits and
bytes are discussed in Chapter 5. A unique number, called a memory address, identifies each storage location. When a CPU performs a memory write, it sends both a
memory address, designating the desired storage location, and the data to be stored
to the memory. When it performs a memory read, it sends the address of the desired
data to the memory and, in return, is sent the data at that location.
One type of memory is commonly called random access memory (RAM). In common usage, RAM refers to memory onto which the CPU can both read and write
data. A better name for such memory is read-write memory. A disadvantage of most
read-write memory is that it is volatile, meaning that data stored in it are lost when
electrical power is turned off.
Another type of memory is called read-only memory (ROM). The CPU can only
read data from ROM; it cannot write or erase data on ROM. The main advantage to
ROM is that data stored in it are not lost when power is lost to the computer. ROM
is usually employed to store frequently used programs provided by the manufacturer that perform important functions such as setting up the computer for operation
when power is turned on. Although the term RAM is commonly used to distinguish
read-write memory from ROM, ROM does permit random access.
The size of the main memory can affect the speed at which a program is executed.
A very large program or one that processes large volumes of information will usually
be executed more quickly on a computer with a large memory. A computer with a
small memory can fit only small portions of the program and information into its
memory at one time and must frequently interrupt the processing of information to
load other portions of the program and information into the memory from the magnetic disk or another storage device.
It would be very expensive to have a large main memory that could provide
program instructions and data to the CPU as fast as the CPU can process them. The
main memories of most computers today use a relatively slow but inexpensive type
of memory called dynamic RAM (DRAM). DRAM uses one capacitor (an electrical
charge storage device) and one transistor for each bit. It is called dynamic because
each bit must be read and refreshed several times a second to maintain the data.
To compensate for the low speed of DRAM with respect to the CPU clock speed,
many computers contain one or more smaller but faster memories, called cache
memories, between the main memory and the CPU. They maintain exact copies
of small portions of the program and data in the main memory. A cache memory
usually consists of static RAM (SRAM), which uses several transistors to represent
each bit. SRAM stores data while power is on without requiring refreshing and can
be read or written several times faster than DRAM but requires more space and is
more expensive. Cache memories are effective because program segments and data
that were recently read from the memory by the CPU are very likely to be read
again soon.
932
Computer Program and Central Processing Unit

A computer program is a sequence of instructions and the CPU is a set of electronic
circuits that executes them. A CPU fetches and executes the instructions in a program sequentially; it fetches an instruction from the main memory, executes that
instruction, and then fetches the next instruction in the program sequence. A CPU
contained in a single computer chip is called a microprocessor.
The CPU contains a small number of data storage locations, called storage registers, for storing data and memory addresses. For example, one computer system
has 16 registers, each containing 32 bits. Operations can be performed on data in
the registers much more quickly than on data in memory. The CPU also contains an
arithmetic logic unit (ALU) that performs mathematical operations and comparisons
between data stored in the registers or memory. In addition, the CPU contains a
clock. This clock is not a time-of-day clock (although the computer may have one
of these as well), but a circuit that produces a periodic stream of logic pulses that
serve as timing marks to synchronize the operation of the various components of the
computer. CPU clock speeds today are usually several gigahertz (GHz); a GHz is 109
hertz (Hz) and a Hz is defined as one cycle per second.
An instruction can cause the CPU to perform one or a combination of four actions:
Transfer a unit of data (e.g., a 16- or 32-bit word) from a memory address, CPU storage register, or I/O device to another memory address, register, or I/O device.
Perform a mathematical operation, such as addition or multiplication, between two
numbers in storage registers in the CPU or in memory.
Compare two numbers or other pieces of data in registers or in memory.
Change the address of the next instruction to be executed in the program to
another location in the program. This type of instruction is called a branching
instruction.
A CPU fetches and executes the instructions in a program sequentially until it
fetches a branching instruction. The CPU then jumps to a new location in the program that is specified by the branching instruction and, starting at the new location,
again executes instructions sequentially. There are two types of branching instructions, unconditional and conditional. Conditional branching instructions cause a branch
to occur only if a specified condition is met. For example, a conditional branching
instruction may specify that a branch is to occur only if the number stored in a particular CPU storage register is zero. Conditional branching instructions are especially
important because they permit the computer to make decisions by allowing the
data to alter the order of instruction execution.
Each instruction in a program has two parts. The first part specifies the operation, such as a data transfer, an addition of two numbers, or a conditional branch, to
be performed. The second part describes the location of the data to be operated on or
compared and the location where the result is to be stored. In the case of an addition,
it contains the locations of the numbers to be added and the locations where the sum
is to be stored. In the case of a branching instruction, the second part is the memory
address of the next instruction to be fetched and executed.
A computer or microprocessor manufacturer creates an architecture for its computers; this term refers to the set of instructions that a CPU can execute and the
formats for data of different types (unsigned integers, signed integers, floating point
numbers, alphanumeric characters, etc.) that the instructions cause the CPU to process. If the computers or microprocessors incorporating an architecture are commercially successful, software companies and individuals write programs for that
933
architecture. When the computer or microprocessor manufacturer creates a faster

CPU or microprocessor (e.g., with a higher clock speed or a design that requires
fewer clock cycles to execute an instruction), if the instruction architecture is not
changed, the new computer or microprocessor will be capable of executing existing programs. A person or an organization may have a large financial investment
in software and store information in that format and will likely prefer to purchase
new computers that can execute this software and access existing data. On the other
hand, such a computer architecture may restrict the ability to take advantage of new
developments in computer technology. A partial solution to this dilemma is to design
new microprocessors that add additional instructions and data formats but continue
to execute all existing instructions and accept existing data formats. Such processors
are said to be backward compatible.
A CPU or microprocessor may be described as being an 8-, 16-, 32-, or 64-bit
design. Modern systems commonly are 32- or 64-bit designs. This terminology is
somewhat ambiguous; the number of bits may refer to the number of bits in each
storage or address register in the CPU, the number of address lines in the data bus,
or the number of data lines in the data bus. For example, if 32 bits are allotted for
address registers and there are 32 address lines in the bus, 4 gigabytes of memory can
be directly accessed. If 32 bits are allotted for storage registers and there are 32 data
lines in the bus, 32 bits of information can be read from or written to memory simultaneously. In general, the number of bits refers to the amount of information that
can be read, written, or processed in a single operation and the number of memory
addresses that can be accessed without schemes like memory segmentation. Thus, a
CPU or microprocessor described as being of a design with more bits will, in general,
process information more quickly, all other factors such as clock speed being equal.
However, this improvement in efficiency requires software that effectively utilizes
these capabilities.
The CPU is a bottleneck in a conventional computer. A computer may have many
megabytes of data to be processed and a program requiring the execution of instructions many billions of times. However, the CPU executes the instructions only one
or a few at a time. One way to increase speed is to build the circuits of the CPU and
memory with faster components so the computer can run at a higher clock speed.
This makes the computer more expensive and there are practical and technical limits
to how fast a circuit can operate.
The other method of relieving the bottleneck is to design the CPU to perform
parallel processing. This means that the CPU performs some tasks simultaneously
rather than sequentially. Most modern general-purpose CPUs incorporate limited
forms of parallel processing. One form of parallel processing is pipelining. Several
steps, each requiring one clock cycle, are required for a CPU to execute a single
instruction. Older CPUs, such as the Intel 8088, 80286, and 80386 microprocessors, completed only one instruction every few cycles. The instruction execution
circuits of newer microprocessors are organized into pipelines, a pipeline having
several stages that operate simultaneously. Instructions enter the pipeline one at
a time and are passed from one stage to the next. Several instructions are in different stages of execution at the same time. It takes several clock cycles to fill the
pipeline, but, once full, it completes the execution of one instruction per cycle.
Pipelining permits a CPU to attain an average instruction execution rate of almost
one instruction per cycle. (Pipelining does not yield exactly one instruction per
cycle because some instructions interfere with its operation, in particular ones that
modify the data to be operated on by immediately subsequent instructions, and
branching instructions, because they change the order of instruction execution.)
934
Superscalar CPU architectures incorporate several pipelines in parallel and can

attain average instruction execution rates of two to five instructions per CPU clock
cycle.
Modern microprocessors may contain more than one CPU. These are called
multicore microprocessors. To effectively use their processing capability, the software
must provide threads of instructions that can be processed simultaneously.
However, many common tasks performed by general-purpose computers are not
well suited to large-scale parallel processing. On the other hand, many tasks in image
processing are very well suited for parallel processing and so some specialized processors for image processing, described later in this chapter, are designed to perform
extensive parallel processing.
Input/Output Bus and Expansion Slots

Data buses have been described earlier in this chapter. The input/output buses of
most computers are provided with several expansion slots into which printed circuit boards can be installed. The boards installed in these slots can include cards to
connect the computer to a computer network, controllers for mass storage devices,
display interface cards to provide a signal to a display monitor, sound cards to provide audio signals to speakers, and acquisition interfaces to other devices such as
scintillation cameras. The provision of expansion slots on this bus makes it possible
to customize a general-purpose computer for a specific application and to add additional functions and capabilities.
Mass Storage Devices

Mass storage devices permit the nonvolatile (i.e., data are not lost when power is turned
off) storage of information. Mass storage devices include hard magnetic disk drives
(nonremovable hard disks are called fixed disks), flash memory, magnetic tape drives,
and optical (laser) disk units. Each of these devices, except flash memory, consists of
a mechanical drive; the storage medium, which may be removable; and an electronic
controller. Despite the wide variety of storage media, there is not one best medium for
all purposes, because they differ in storage capacity, data access time, data transfer rate,
cost, and other factors. Mass storage devices were discussed in detail in Chapter 5.
Display Interface
An important use of many computers, particularly workstations, is to display information in visual form. The information is usually displayed on one or more display
monitors, although it may also be presented in hard-copy form, most commonly
using photographic film or paper. Such a computer has a display interface; the display interface is a collection of electronic circuits that receives information in digital
form from the memory of a computer via the computers bus and provides it, in analog or digital form, to one or more display monitors. Display interfaces and display
devices are described in detail in Chapter 5.
Acquisition Interface
A computers acquisition interface accepts data from an attached device and transfers
it to memory via the computers bus. An acquisition interface that accepts analog
data first converts it into digital form using one or more analog-to-digital converters
935
(ADCs) (ADCs were described in Chapter 5.). Until perhaps a decade ago, many
medical imaging devices, such as scintillation cameras, sent analog data directly to
attached computers. In most cases today, the analog data are converted into digital
form by the imaging device, which sends data in digital form to the computer. Some
acquisition interfaces store the received information in their own dedicated memories to permit the computer to perform other tasks during data acquisition. An imaging computer that is not directly connected to an imaging device may instead receive
images over a computer network, as described below.
Communications Interfaces
Computers serve not only as information processing and display devices but also as
communications devices. A computer may be connected to a network by a network
interface. The interface converts digital signals from the computer into the digital
form required by the network and vice versa. Various forms of the local area network
standard Ethernet have become so common that most computers today are equipped
with Ethernet interfaces. There are also wireless interfaces, such as WiFi, which use
radio signals to connect portable computers to a network. For example, in bedside
radiography, such wireless network technology may be used to transfer information
between a portable image receptor and a mobile radiographic system. A computer
may have a modem (contraction of modulator-demodulator) to permit it to transfer
information to another computer or to a network using analog telephone lines. These
interfaces and computer networks are discussed in detail in Chapter 5.
Specialized Processors
The speed of a general-purpose computer in performing mathematical operations
can be greatly enhanced by adding a specialized processor. A specialized processor is a specialized computer designed to perform the mathematical manipulation
of data very quickly. It typically consists of a CPU, a small high-speed memory, a
high-speed bus, and an interface to connect it to the bus of the main computer. The
main computer is called the host computer. Most specialized processors rely heavily
on parallel processing for their speed. Parallel processing is particularly efficient for
many image processing tasks. The host computer sends the data from its memory
to the specialized processors memory along with the instructions for processing the
information. The specialized processor processes the data and returns the processed
information to the memory of the host computer. Alternatively, digital information
from an imaging device, such as the digital projection data from a CT scanner, may
be sent directly to a specialized processor, with the processed information, such as
transverse images, sent to the host computer. Figure B-2 shows a computer with a
specialized processor. Uses of specialized processors in medical imaging include the
production of tomographic image sets from projection images in CT, SPECT, and
PET and the production of surface- and volume-rendered images from tomographic
image data sets. Specialized processors for image processing are commonly used on
display cards; these are known as graphics processing units (GPUs).
B.2 Performance of Computers

There are many factors that affect the time required for a computer to complete a
task. These include the speed of the CPU (as reflected by its clock speed); how extensively the CPU uses parallel processing; the number of independent CPUs, called
936
FIGURE B-2 Specialized high-capacity

parallel processor (arrows) for reconstructing tomographic images from projection
image data for a modern multirow detector CT system. The parallel processor is
mounted above the motherboard of a very
powerful, but conventional personal computer that receives the image data from
the specialized processor. (Image courtesy
of Philips Healthcare.)
cores, that a microprocessor contains (four or more cores are now typical); the
width and the clock speed of the data bus between the CPU and the main memory;
the size of the memory; and the access times and data transfer rates of mass storage
devices. A larger memory permits more of the program being executed and the data
being processed to reside in memory, thereby reducing the time that the CPU is idle
while awaiting the transfer of program segments and data from mass storage devices
into memory. The widths (number of data wires) of the data buses affect the speed
of the computer; a 32-bit bus can transfer twice as much data per clock cycle as a
16-bit bus. A program that is multithreaded divides a processing task into separate
threads of instructions that are sent to individual cores in a multicore microprocessor, thereby distributing individual tasks for simultaneous processing. Multitasking
(running more than one process at one time) can also take advantage of multiple core
microprocessor designs, thus enhancing the data and processing throughput.
There are several ways to compare the performance of computer systems. One
indicator of CPU speed is the speed of its clock. If two computers are otherwise identical, the computer with the faster clock speed is faster. For example, if the CPUs of
two computers are otherwise identical, a CPU running at 2 gigahertz (GHz) is faster
than a CPU running at 1 GHz.
The clock speed of a CPU is not as useful in comparing computers with different microprocessors, because one microprocessor architecture may execute more
instructions per clock cycle than another. To avoid this problem, the speed of a CPU
may be measured in millions of instructions per second (MIPS). If a computer is used to
performing extensive mathematical operations, it may be better to measure its performance in millions or billions of floating-point operations per second (megaflops
or gigaflops). Another way to compare the performance of computers is to compare
the times it takes them to execute standard programs. Programs used for this purpose
are called benchmarks.
The ability of a computer to execute a particular program in less time than another
computer or to perform a greater number of MIPS or megaflops does not prove that
it is faster for all applications. For example, the execution speed of one application
may rely primarily on CPU speed and CPU-memory transfer rates, whereas another
applications speed of execution may largely depend on the rate of transfer of data
between the hard disk and main memory. When evaluating computers for a particular purpose, such as reconstructing tomographic images, it is better to compare their
speed in performing the actual tasks expected of them than to compare them solely
on the basis of clock speed, MIPS, megaflops, or benchmark performance.
937
B.3 Computer Software
Computer Languages
Machine Language
The actual binary instructions that are executed by the CPU are written in machine
language. There are disadvantages to writing a program in machine language. The
programmer must have a very detailed knowledge of the computers construction to
write such a program. Also, machine-language programs are very difficult to read and
understand. For these reasons, programs are seldom written in machine language
today
High-Level Languages
To make programming easier and to make programs more readable, high-level languages, such as Fortran, Basic, Pascal, C, and Java, have been developed. These languages enable computer programs to be written using statements similar to simple
English, resulting in programs that are much easier to read than those in machine
language. Also, much less knowledge of the particular computer is required to write
a program in a high-level language. Using a program written in a high-level language
requires a program called a compiler or an interpreter to convert the statements written in the high-level language into machine-language instructions that the CPU can
execute.
Applications Programs
An applications program, commonly referred to as an application, is a program that
performs a specific function for a user. Examples of applications are e-mail programs,
word processing programs, web browsers, and image display programs.
Operating System
Even when a computer is not executing an assigned program and seems to be idle,
there is a fundamental program, called the operating system, being executed. When
the user instructs a computer to run a particular program, the operating system copies the program into memory from a disk, transfers control to it, and regains control
of the computer when the program has finished. The operating system also handles
most of the complex aspects of input/output operations and controls the sharing of
resources on multiuser computers. Common operating systems include Windows
and Linux for IBM-compatible PCs; the Mac OS X on Apple computers; and Unix,
used on a wide variety of computers.
Most operating systems store information on mass storage devices as files. A single file may contain a program, one or more images, patient data, or a manuscript.
Each storage device or unit of storage media has a directory that lists all files on the
device and describes the location of each.
The operating system can be used to perform many utility functions. These utility
functions include listing the files on a mass storage device, determining how much
unused space remains on a mass storage device, deleting files, and making a copy of
a file on another device.
APPENDIX
Physical Constants, Prefixes,

Geometry, Conversion Factors,
and Radiologic Data
TABLE C-1 PHYSICAL CONSTANTS, PREFIXES,
AND GEOMETRY
A: Physical Constants
Acceleration due to gravity
9.80665 m/s2
Gyromagnetic ratio of proton
42.5764 MHz/T
Avogadros number
6.0221415 3 1023 particles/mole
Plancks constant
6.62607 3 10234 J-s
Plancks constant
4.13567 3 10215 eV-s
Charge of electron
1.602177 3 10219 C
Mass of electron
9.109382 3 10231 kg
Mass of proton
1.672623 3 10227 kg
Mass of neutron
1.674929 3 10227 kg
Atomic mass unit
1.660539 3 10227 kg
Molar volume
22.465 L/mol (at 25C)
Speed of light (c)
2.99792458 3 108 m/s
B: Prefixes
yotta (Y)
1024
zetta (Z)
1021
exa (E)
1018
peta (P)
1015
tera (T)
1012
giga (G)
109
mega (M)
106
kilo (k)
103
hecto (h)
102
deca (da)
101
deci (d)
1021
centi (c)
1022
(Continued)
938
Appendix C Physical Constants, Prefixes, Geometry, Conversion Factors, Radiologic Data 939
TABLE C-1 PHYSICAL CONSTANTS, PREFIXES,
AND GEOMETRY (Continued)

milli (m)
1023
micro ()
1026
nano (n)
1029
pico (p)
10212
femto (f)
10215
atto (a)
10218
zepto (z)
10221
yocto (y)
10224
C: Geometry
Area of circle
pr2
Circumference of circle
2 pr
Surface area of sphere
4 pr2
Volume of sphere
(4/3) pr3
Area of triangle
(1/2) base 3 height
TABLE C-2 Definitions and conversion factors

Energy
1 electron volt (eV)
1.6022 3 10219 J
1 calorie
4.187 J
Power
1 watt (W)
J/s
Current
1 ampere
1 C/s
1 ampere
6.241 3 1018 electrons/s
Volume
1 US gallon
3.7854 L
1 liter (L)
1,000 cm3
Temperature
C (Celsius)
5/9 3 (F 2 32)
F (Fahrenheit)
9/5 3 C 1 32
K (Kelvin)
C 1 273.16
Mass
1 pound
0.45359 kg
1 kg
2.2 pound
Length
1 inch
25.4 mm
1 inch
2.54 cm
(Continued)
940
TABLE C-2DEFINITIONS AND CONVERSION FACTORS
(Continued)
Angle
1 radian
360/2p 57.2958 degrees
Radiologic Units
1 roentgen (R)
2.5 3 10-4 C/kg
1 roentgen (R)
8.73 mGy air kerma
1 mGy air kerma
114.5 mR
1 gray (Gy)
100 rad
1 sievert (Sv)
100 rem
1 curie (Ci)
3.7 3 1010 becquerel (Bq)
1 becquerel (Bq)
1 disintegration s1 (dps)
Li
Be
Ne
Na
Mg
Al
Si
Cl
Ar
Ca
Sc
10
11
12
13
14
15
16
17
18
19
20
21
He
SYM
Scandium
Calcium
Potassium
Argon
Chlorine
Sulfur
Phosphorus
Silicon
Aluminum
Magnesium
Sodium
Neon
Fluorine
Oxygen
Nitrogen
Carbon
Boron
Beryllium
Lithium
Helium
Hydrogen
Element
2.989
1.54
0.862
0.0017837
0.003214
2.067
1.82
2.3296
2.698
1.738
0.971
0.0008999
0.001696
0.001429
0.0012506
2.267
2.34
1.85
0.534
0.0001785
8.988E-05
DENSITY
(g/cm3)
44.958
40.08
39.102
39.948
35.453
32.064
30.974
28.086
26.982
24.312
22.99
20.179
18.998
15.999
14.007
12.011
10.811
9.012
6.939
4.003
1.008
AT MASS
(g/mol)
4.965
4.486
4.034
3.607
3.203
2.824
2.471
2.143
1.840
1.309
1.080
0.874
0.692
0.533
0.397
0.282
0.188
0.115
0.0550
0.0246
0.0136
K-EDGE
(kev)
0.529
0.462
0.403
0.340
0.286
0.237
0.193
0.153
0.118
0.062
0.06
L-edges:
LI,(keV)
TABLE C-3 RADIOLOGIC DATA FOR ELEMENTS 1 THROUGH 100
0.400
0.346
0.296
0.247
0.204
0.164
0.130
0.101
0.076
0.08
L-edges
LII (keV)
0.396
0.342
0.293
0.245
0.202
0.163
0.129
0.100
0.075
0.056
0.039
0.026
0.015
0.008
0.004
0.005
0.005
0.006
L-edges:
LIII (keV)
4.090
3.692
3.314
2.958
2.622
2.308
2.014
1.740
1.487
1.253
1.041
0.848
0.677
0.525
0.393
0.277
0.183
0.109
0.052
Ka1 (keV)
4.086
3.688
3.311
2.956
2.620
2.307
2.013
1.739
1.486
1.253
1.041
0.848
0.677
0.525
0.393
0.277
0.183
0.109
0.052
Ka2 (keV)
4.461
4.013
3.59
3.19
2.816
2.464
2.139
1.836
1.557
1.302
1.071
0.858
Kb1 (keV)
(Continued)
4.461
4.013
3.59
3.19
2.816
2.464
2.139
1.836
1.557
1.302
1.071
0.858
Kb2 (keV)
Molybdenum
Technetium
Cr
Mn
Fe
Co
Ni
Cu
Zn
Ga
Ge
As
Se
Br
Kr
Rb
Sr
Zr
Nb
Mo
Tc
Ru
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Ruthenium
Niobium
Zirconium
Yttrium
Strontium
Rubidium
Krypton
Bromine
Selenium
Arsenic
Germanium
Gallium
Zinc
Copper
Nickel
Cobalt
Iron
Manganese
Chromium
Vanadium
Titanium
Ti
22
Element
SYM
12.37
11.5
10.22
8.57
6.506
4.469
2.64
1.532
0.003733
3.122
4.809
5.776
5.323
5.907
7.134
8.96
8.912
8.86
7.874
7.44
7.15
6.11
4.54
DENSITY
(g/cm3)
101.7
99
95.94
92.906
91.22
88.905
87.62
85.47
83.8
79.909
78.96
74.922
72.59
69.72
65.37
63.54
58.71
58.933
55.847
54.938
51.996
50.942
47.9
AT MASS
(g/mol)
23.229
22.123
21.048
20.002
18.985
17.997
17.037
16.106
15.202
14.330
13.483
12.666
11.877
11.115
10.386
9.673
8.993
8.339
7.712
7.112
6.537
5.987
5.463
K-EDGE
(kev)
3.421
3.230
3.047
2.867
2.698
2.535
2.372
2.217
2.067
1.923
1.791
1.662
1.536
1.426
1.316
1.208
1.109
1.016
0.929
0.846
0.768
0.694
0.626
L-edges:
LI,(keV)
2.973
2.797
2.628
2.464
2.305
2.155
2.008
1.866
1.732
1.605
1.485
1.368
1.259
1.155
1.057
0.965
0.878
0.797
0.721
0.649
0.582
0.519
0.460
L-edges
LII (keV)
TABLE C-3 RADIOLOGIC DATA FOR ELEMENTS 1 THROUGH 100 (Continued)

(Continued)
2.844
2.681
2.523
2.370
2.227
2.079
1.940
1.806
1.680
1.559
1.444
1.333
1.228
1.134
1.034
0.945
0.861
0.782
0.708
0.638
0.572
0.511
0.454
L-edges:
LIII (keV)
19.279
18.367
17.479
16.615
15.770
14.958
14.166
13.396
12.650
11.924
11.222
10.544
9.887
9.252
8.639
8.048
7.478
6.930
6.404
5.899
5.415
4.952
4.511
Ka1 (keV)
Ka2 (keV)
19.150
18.251
17.374
16.521
15.692
14.882
14.098
13.336
12.598
11.878
11.181
10.509
9.856
9.231
8.616
8.028
7.461
6.915
6.391
5.888
5.405
4.944
4.505
21.656
20.619
19.608
18.623
17.662
16.737
15.836
14.962
14.113
13.292
12.496
11.727
10.983
10.271
9.572
8.905
8.265
7.649
7.058
6.49
5.947
5.427
4.932
Kb1 (keV)
22.072
21.002
19.960
18.9471
17.9631
17.0151
16.0851
15.1851
14.3151
13.4701
12.6521
11.8641
11.1011
10.3661
9.6581
8.93
8.265
7.649
7.058
6.49
5.947
5.427
4.932
Kb2 (keV)
942
Rh
Pd
Ag
Cd
In
Sn
Sb
Te
Xe
Cs
Ba
La
Ce
Pr
Nd
Pm
Sm
Eu
Gd
Tb
Dy
Ho
Er
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
Erbium
Holmium
Dysprosium
Terbium
Gadolinium
Europium
Samarium
Promethium
Neodymium
Praseodymium
Cerium
Lanthanum
Barium
Cesium
Xenon
Iodine
Tellurium
Antimony
Tin
Indium
Cadmium
Silver
Palladium
Rhodium
9.066
8.795
8.55
8.229
7.895
5.243
7.52
7.26
7.007
6.773
6.77
6.145
3.594
1.873
0.005887
4.93
6.232
6.685
7.287
7.31
8.69
10.501
12.02
12.41
167.26
164.93
162.5
158.924
157.25
151.96
150.35
145
144.24
140.907
140.12
138.91
137.34
132.905
131.3
126.904
127.6
121.75
118.69
114.82
112.4
107.87
106.4
102.905
61.322
59.374
57.464
55.593
53.761
51.965
50.207
48.486
46.801
45.152
43.538
41.958
40.410
38.894
35.966
34.551
33.168
31.817
30.499
29.211
27.953
26.727
25.531
24.365
10.479
10.097
9.725
9.365
9.013
8.679
8.343
8.024
7.707
7.398
7.095
6.802
6.516
6.235
5.700
5.442
5.186
4.942
4.706
4.475
4.250
4.034
3.822
3.619
9.243
8.894
8.553
8.221
7.898
7.584
7.278
6.981
6.691
6.408
6.131
5.860
5.597
5.341
5.092
4.851
4.616
4.389
4.167
3.951
3.742
3.540
3.344
3.156
8.336
8.046
7.762
7.484
7.211
6.944
6.683
6.427
6.177
5.932
5.690
5.452
5.220
4.993
4.772
4.556
4.345
4.140
3.939
3.744
3.554
3.368
3.187
3.013
49.128
47.547
45.999
44.481
42.996
41.542
40.118
38.725
37.361
36.026
34.720
33.442
32.194
30.973
29.779
28.612
27.472
26.359
25.272
24.209
23.173
22.163
21.178
20.216
48.221
46.699
45.208
43.744
42.309
40.902
39.523
38.171
36.847
35.550
34.279
33.034
31.817
30.625
29.459
28.317
27.201
26.110
25.044
24.002
22.985
21.991
21.021
20.073
55.681
53.878
52.119
50.382
48.696
47.036
45.413
43.825
42.272
40.748
39.258
37.802
36.378
34.985
33.625
32.295
30.995
29.725
28.491
27.275
26.095
24.943
23.819
22.723
(Continued)
57.164
55.308
53.491
51.709
49.964
48.256
46.584
44.947
43.345
41.778
40.243
38.739
37.270
35.833
34.428
33.054
31.712
30.402
29.122
27.872
26.653
25.463
24.303
23.173
SYM
Tm
Yb
Lu
Hf
Ta
Re
Os
Ir
Pt
Au
Hg
Tl
Pb
Bi
Po
At
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
Astatine-210
Polonium-209
Bismuth
Lead
Thallium
Mercury
Gold
Platinum
Iridium
Osmium
Rhenium
Tungsten
Tantalum
Hafnium
Lutetium
Ytterbium
Thulium
Element
9.32
9.807
11.342
11.85
13.5336
19.282
21.46
22.56
22.61
21.02
19.25
16.654
13.31
9.84
6.965
9.321
DENSITY
(g/cm3)
209.983
208.98
208.98
207.19
204.37
200.5
196.967
195.09
192.2
190.2
186.2
183.85
180.948
178.49
174.97
173.04
168.934
AT MASS
(g/mol)
101.131
98.402
95.730
93.113
90.540
88.008
83.532
83.109
80.729
78.400
76.111
73.869
71.670
69.517
67.405
65.345
63.311
K-EDGE
(kev)
18.630
18.047
17.495
16.940
16.391
15.863
15.344
14.835
14.353
13.880
13.416
12.968
12.522
12.092
11.672
11.262
10.869
L-edges:
LI,(keV)
16.787
16.250
15.725
15.204
14.700
14.215
13.739
13.277
12.824
12.382
11.952
11.535
11.128
10.734
10.346
9.968
9.601
L-edges
LII (keV)
TABLE C-3 RADIOLOGIC DATA FOR ELEMENTS 1 THROUGH 100 (Continued)

(Continued)
14.207
13.812
13.422
13.039
12.660
12.290
11.925
11.568
11.215
10.868
10.530
10.199
9.872
9.555
9.241
8.933
8.632
L-edges:
LIII (keV)
81.523
79.301
77.118
74.969
72.872
70.819
68.804
66.832
64.896
63.001
61.140
59.318
57.533
55.790
54.070
52.389
50.742
Ka1 (keV)
78.943
76.863
74.815
72.804
70.832
68.894
66.990
65.123
63.287
61.487
59.718
57.982
56.277
54.611
52.965
51.354
49.773
Ka2 (keV)
92.302
89.801
87.354
84.936
82.575
80.261
77.985
75.751
73.560
71.416
69.309
67.244
65.221
63.236
61.286
59.374
57.513
Kb1 (keV)
94.983
92.403
89.866
87.367
84.924
82.532
80.182
77.883
75.620
73.404
71.230
69.100
67.011
64.973
62.960
60.991
59.059
Kb2 (keV)
944
Th
Pa
90
91
92
Bk
Cf
Es
Fm
97
98
99
100
Am
Cm
95
96
Np
Ac
89
Pu
Ra
88
93
Fr
87
94
Rn
86
Fermium-257
Einsteinium-252
Californium-251
Berkelium-247
Curium-247
Americium-241
Plutonium-239
Neptunium-237
Uranium-238
Protactinium-231
Thorium-232
Actinium-227
Radium-226
Francium-223
Radon-222
13.5
15.1
14.79
13.51
13.69
19.84
20.45
18.95
15.37
11.72
10.07
5.5
1.87
0.00973
257.095
252.083
251.08
247.07
247.07
241.05
239.052
237.048
238.051
231.036
232.038
227.028
226.025
223.02
222.018
150.540
146.780
143.090
139.490
135.960
131.590
128.220
125.027
121.818
118.678
115.606
112.599
109.641
106.738
103.909
29.380
28.530
27.700
26.900
26.110
25.275
24.460
23.773
23.097
22.427
21.757
21.088
20.449
19.823
19.222
26.810
26.020
25.250
24.385
23.779
22.944
22.266
21.601
20.947
20.312
16.689
19.063
18.475
17.900
17.337
20.900
20.410
19.930
19.452
18.930
18.504
18.057
17.610
17.166
16.716
16.283
15.854
15.433
15.017
14.609
122.190
119.080
116.030
112.138
109.290
106.523
103.761
101.068
98.440
95.883
93.358
90.884
88.476
86.114
83.793
116.280
113.470
110.710
107.205
104.441
102.083
99.552
97.077
94.659
92.287
89.952
87.675
85.434
83.231
81.065
137.693
134.238
130.851
126.663
123.423
120.360
117.261
114.243
111.303
108.435
105.611
102.846
100.130
97.477
94.866
141.724
138.169
134.681
130.355
127.066
123.891
120.703
117.591
114.561
111.606
108.690
105.837
103.039
100.306
97.617
946
7.83E102
3.58E102
1.92E102
1.15E102
7.37E101
5.04E101
3.56E101
2.60E101
1.53E101
9.73E100
6.58E100
4.65E100
3.42E100
1.83E100
1.13E100
7.69E201
5.67E201
10
12
14
16
18
20
25
30
35
40
5 2.699
1.83E100
2.63E100
4.07E100
6.88E100
1.30E101
1.76E101
2.48E101
3.62E101
5.60E101
9.30E101
1.25E102
1.72E102
2.46E102
3.74E102
5.97E102
1.20E102
2.71E102
5 1.55
Z 5 20
Z 5 13
Energy (keV)
CALCIUM
ALUMINUM
4.85E100
7.05E100
1.09E101
1.82E101
3.36E101
4.49E101
6.20E101
8.91E101
1.35E102
2.14E102
2.76E102
5.18E101
7.48E101
1.15E102
1.89E102
3.46E102
7.47E102
5 8.96
Z 5 29
COPPER
1.29E101
1.85E101
2.82E101
4.78E101
7.89E101
1.72E101
2.37E101
3.40E101
5.22E101
8.55E101
1.13E102
1.56E102
2.23E102
3.38E102
5.42E102
9.65E102
2.01E103
5 10.22
Z 5 42
MOLYBDENUM
1.54E101
2.21E101
3.32E101
5.29E101
1.57E101
2.10E101
2.92E101
4.21E101
6.44E101
1.05E102
1.39E102
1.90E102
2.72E102
4.11E102
6.55E102
1.17E103
4.42E102
5 12.41
Z 5 45
RHODIUM
2.20E101
3.10E101
8.38E100
1.37E101
2.52E101
3.35E101
4.61E101
6.61E101
1.00E102
1.62E102
2.13E102
2.89E102
4.13E102
6.10E102
8.30E102
3.56E102
7.49E102
5 4.93
Z 5 53
IODINE
1.06E101
1.51E101
2.28E101
1.43E101
2.02E101
3.06E101
4.93E101
8.61E101
3.66E101
1.10E102
6.56E101
1.51E102
1.33E102
8.15E101
1.29E102
1.68E102
2.27E102
3.16E102
4.73E102
7.28E102
1.25E103
1.96E103
5 11.35
Z 5 82
LEAD
8.63E101
1.17E102
1.65E102
2.10E102
9.33E101
1.24E102
1.69E102
2.36E102
3.55E102
5.51E102
9.54E102
1.92E103
5 19.3
Z 5 74
TUNGSTEN
TABLE D-1 MASS ATTENUATION COEFFICIENTS IN cm2/g (DENSITY () IN g/cm3)
APPENDIX
Mass Attenuation Coefficients
D.1 Mass Attenuation Coefficients for Selected Elements
4.46E201
3.68E201
3.15E201
2.78E201
2.52E201
2.30E201
2.14E201
2.02E201
1.92E201
1.83E201
1.77E201
1.71E201
1.61E201
1.53E201
1.47E201
1.42E201
1.38E201
1.34E201
1.31E201
45
50
55
60
65
70
75
80
85
90
95
100
110
120
130
140
150
160
170
1.53E201
1.59E201
1.67E201
1.77E201
1.90E201
2.06E201
2.28E201
2.58E201
2.79E201
3.00E201
3.33E201
3.66E201
4.12E201
4.72E201
5.56E201
6.61E201
8.07E201
1.02E100
1.34E100
1.89E201
2.03E201
2.23E201
2.47E201
2.81E201
3.23E201
3.78E201
4.60E201
5.20E201
5.80E201
6.71E201
7.63E201
9.00E201
1.06E100
1.30E100
1.60E100
2.01E100
2.62E100
3.51E100
3.31E201
3.69E201
4.23E201
4.91E201
5.88E201
7.10E201
8.66E201
1.10E100
1.27E100
1.44E100
1.70E100
1.96E100
2.33E100
2.80E100
3.46E100
4.29E100
5.42E100
7.07E100
9.38E100
3.82E201
4.29E201
4.94E201
5.77E201
6.95E201
8.43E201
1.04E100
1.32E100
1.53E100
1.74E100
2.05E100
2.37E100
2.84E100
3.39E100
4.20E100
5.18E100
6.52E100
8.42E100
1.12E101
5.31E201
6.01E201
7.00E201
8.26E201
1.00E100
1.22E100
1.52E100
1.95E100
2.26E100
2.57E100
3.05E100
3.53E100
4.19E100
5.02E100
6.18E100
7.61E100
9.56E100
1.23E101
1.62E101
1.18E100
1.36E100
1.59E100
1.89E100
2.29E100
2.83E100
3.47E100
4.42E100
5.10E100
5.78E100
6.78E100
7.80E100
9.21E100
1.09E101
2.92E100
3.62E100
4.54E100
5.84E100
7.72E100
(Continued)
1.50E100
1.72E100
2.02E100
2.38E100
2.90E100
3.55E100
4.36E100
5.54E100
6.38E100
7.20E100
1.98E100
2.32E100
2.77E100
3.30E100
4.04E100
4.98E100
6.20E100
7.98E100
1.05E101

Appendix D Mass Attenuation Coefficients
947
9.76E202
9.26E202
8.82E202
8.43E202
8.10E202
7.80E202
7.52E202
350
400
450
500
550
600
650
6.83E202
1.04E201
300
800
1.12E201
250
7.27E202
1.22E201
200
7.05E202
1.25E201
190
750
1.27E201
180
700
5 2.699
7.11E202
7.34E202
7.56E202
7.83E202
8.13E202
8.45E202
8.83E202
9.26E202
9.75E202
1.04E201
1.11E201
1.22E201
1.37E201
1.42E201
1.47E201
5 1.55
Z 5 20
Z 5 13
Energy (keV)
CALCIUM
ALUMINUM
6.59E202
6.81E202
7.04E202
7.30E202
7.61E202
7.94E202
8.35E202
8.82E202
9.39E202
1.01E201
1.12E201
1.28E201
1.56E201
1.65E201
1.75E201
5 8.96
Z 5 29
COPPER
6.61E202
6.86E202
7.13E202
7.45E202
7.82E202
8.27E202
8.83E202
9.52E202
1.04E201
1.18E201
1.38E201
1.73E201
2.43E201
2.68E201
2.93E201
5 10.22
Z 5 42
MOLYBDENUM
6.70E202
6.97E202
7.25E202
7.60E202
8.02E202
8.49E202
9.11E202
9.90E202
1.10E201
1.25E201
1.49E201
1.91E201
2.75E201
3.06E201
3.36E201
5 12.41
Z 5 45
RHODIUM
6.73E202
7.05E202
7.37E202
7.79E202
8.31E202
8.89E202
9.67E202
1.07E201
1.22E201
1.43E201
1.80E201
2.40E201
3.68E201
4.14E201
4.61E201
5 4.93
Z 5 53
IODINE
1.02E201
8.02E202
8.82E202
9.53E202
9.16E202
8.59E202
1.13E201
1.26E201
1.40E201
1.61E201
1.90E201
2.33E201
2.97E201
4.07E201
6.02E201
1.00E100
1.15E100
1.29E100
5 11.35
Z 5 82
LEAD
9.98E202
1.10E201
1.21E201
1.37E201
1.60E201
1.92E201
2.43E201
3.26E201
4.75E201
7.90E201
9.03E201
1.02E100
5 19.3
Z 5 74
TUNGSTEN
TABLE D-1 MASS ATTENUATION COEFFICIENTS IN cm2/g (DENSITY () IN g/cm3) (Continued)
948
5.20E102
1.63E102
7.43E101
3.84E101
2.24E101
1.40E101
9.42E100
6.62E100
4.84E100
2.87E100
1.85E100
1.29E100
9.55E201
7.43E201
4.66E201
3.46E201
2.80E201
2.44E201
2.21E201
10
12
14
16
18
20
25
30
35
40
45
2.44E201
2.68E201
3.07E201
3.75E201
5.09E201
8.08E201
1.04E100
1.40E100
2.02E100
3.16E100
5.30E100
7.23E100
1.03E101
1.54E101
2.47E101
4.24E101
8.24E101
1.95E102
6.17E102
5 1.00
5 0.001293
ENERGY (keV)
WATER
AIR
2.19E201
2.34E201
2.60E201
3.04E201
3.86E201
5.70E201
7.12E201
9.36E201
1.31E100
2.01E100
3.34E100
4.54E100
6.46E100
9.66E100
1.54E101
2.66E101
5.23E101
1.26E102
4.04E102
5 1.19
PLEXIGLASa
2.44E201
2.69E201
3.09E201
3.79E201
5.16E201
8.23E201
1.06E100
1.43E100
2.05E100
3.21E100
5.37E100
7.32E100
1.04E101
1.55E101
2.47E101
4.22E101
8.16E101
1.87E102
5.67E102
5 1.06
MUSCLE
4.18E201
5.20E201
6.86E201
9.80E201
1.56E100
2.82E100
3.78E100
5.26E100
7.67E100
1.19E101
1.99E101
2.68E101
3.71E101
5.36E101
8.26E101
1.34E102
1.06E102
2.42E102
5.24E102
5 1.5 TO 3.0)
BONE
TABLE D-2 MASS ATTENUATION COEFFICIENTS IN cm2/g (DENSITY () IN g/cm3)
D.2 Mass Attenuation Coefficients for Selected Compounds
2.21E201
2.35E201
2.59E201
3.00E201
3.74E201
5.42E201
6.71E201
8.74E201
1.22E100
1.85E100
3.06E100
4.15E100
5.90E100
8.81E100
1.41E101
2.43E101
4.77E101
1.15E102
3.70E102
5 0.930
ADIPOSEb
2.28E201
2.46E201
2.75E201
3.26E201
4.22E201
6.37E201
8.03E201
1.07E100
1.51E100
2.33E100
3.87E100
5.28E100
7.50E100
1.12E101
1.79E101
3.09E101
6.04E101
1.44E102
4.60E102
5 0.982
50/50b
(Continued)
2.34E201
2.55E201
2.90E201
3.49E201
4.64E201
7.23E201
9.22E201
1.24E100
1.77E100
2.75E100
4.60E100
6.28E100
8.94E100
1.34E101
2.14E101
3.68E101
7.17E101
1.71E102
5.41E102
5 1.040
GLANDULARb

949
1.94E201
1.86E201
1.80E201
1.74E201
1.69E201
1.65E201
1.62E201
1.59E201
1.56E201
1.54E201
1.49E201
1.45E201
1.42E201
1.38E201
1.35E201
1.33E201
1.30E201
1.27E201
1.25E201
1.23E201
55
60
65
70
75
80
85
90
95
100
110
120
130
140
150
160
170
180
190
200
1.37E201
1.39E201
1.42E201
1.44E201
1.47E201
1.50E201
1.54E201
1.57E201
1.61E201
1.65E201
1.71E201
1.73E201
1.76E201
1.79E201
1.83E201
1.88E201
1.93E201
1.99E201
2.06E201
2.14E201
2.27E201
5 1.00
2.05E201
5 0.001293
50
ENERGY (keV)
WATER
AIR
1.32E201
1.35E201
1.37E201
1.40E201
1.43E201
1.46E201
1.49E201
1.52E201
1.55E201
1.59E201
1.64E201
1.66E201
1.69E201
1.72E201
1.75E201
1.78E201
1.82E201
1.87E201
1.92E201
1.98E201
2.07E201
5 1.19
PLEXIGLASa
1.35E201
1.38E201
1.40E201
1.43E201
1.46E201
1.49E201
1.52E201
1.33E201
1.36E201
1.38E201
1.42E201
1.45E201
1.49E201
1.53E201
1.64E201
1.58E201
1.56E201
1.71E201
1.80E201
1.86E201
1.92E201
2.00E201
2.08E201
2.20E201
2.34E201
2.52E201
2.75E201
3.07E201
3.53E201
5 1.5 TO 3.0)
BONE
1.60E201
1.64E201
1.69E201
1.72E201
1.75E201
1.78E201
1.82E201
1.86E201
1.92E201
1.98E201
2.05E201
2.14E201
2.26E201
5 1.06
MUSCLE
TABLE D-2 MASS ATTENUATION COEFFICIENTS IN cm2/g (DENSITY () IN g/cm3) (Continued)
1.36E201
1.39E201
1.41E201
1.44E201
1.46E201
1.49E201
1.52E201
1.56E201
1.60E201
1.63E201
1.68E201
1.70E201
1.73E201
1.76E201
1.79E201
1.82E201
1.86E201
1.91E201
1.96E201
2.02E201
2.10E201
5 0.930
ADIPOSEb
1.35E201
1.38E201
1.40E201
1.43E201
1.46E201
1.49E201
1.52E201
1.55E201
1.59E201
1.63E201
1.68E201
1.70E201
1.73E201
1.76E201
1.79E201
1.83E201
1.87E201
1.92E201
1.98E201
2.05E201
2.15E201
5 0.982
50/50b
1.35E201
1.37E201
1.40E201
1.42E201
1.45E201
1.48E201
1.51E201
1.55E201
1.59E201
1.63E201
1.68E201
1.70E201
1.73E201
1.76E201
1.80E201
1.84E201
1.88E201
1.94E201
2.00E201
2.08E201
2.19E201
5 1.040
GLANDULARb
950
1.07E201
1.00E201
9.53E202
9.09E202
8.70E202
8.36E202
8.05E202
7.77E202
7.51E202
7.29E202
7.06E202
300
350
400
450
500
550
600
650
700
750
800
7.85E202
8.10E202
8.35E202
8.64E202
8.96E202
9.29E202
9.67E202
1.01E201
1.06E201
1.11E201
1.18E201
1.27E201
7.63E202
7.87E202
8.11E202
8.39E202
8.70E202
9.03E202
9.39E202
9.82E202
1.03E201
1.08E201
1.15E201
1.23E201
7.78E202
8.03E202
8.27E202
8.56E202
8.87E202
9.21E202
9.58E202
1.00E201
1.05E201
1.10E201
1.17E201
1.25E201
7.51E202
7.75E202
7.98E202
8.26E202
8.56E202
8.89E202
9.26E202
9.68E202
1.02E201
1.07E201
1.14E201
1.22E201
7.84E202
8.09E202
8.34E202
8.62E202
8.94E202
9.27E202
9.65E202
1.01E201
1.06E201
1.11E201
1.18E201
1.26E201
7.79E202
8.04E202
8.29E202
8.57E202
8.89E202
9.22E202
9.60E202
1.00E201
1.05E201
1.11E201
1.17E201
1.25E201
7.75E202
7.99E202
8.24E202
8.52E202
8.83E202
9.17E202
9.54E202
9.97E202
1.04E201
1.10E201
1.17E201
1.25E201
Polymethyl methacrylate, C5H8O2.

Composition data from Hammerstein GR, Miller DW, White DR, et al. Absorbed radiation dose in mammography. Radiology 1979;130:485491. The 50% adipose data is presented here for reference.
More recent information suggests that the typical breast is not 50%250%, but much lower in terms of volume glandular fraction. See: MJ Yaffe, JM Boone, N Packard, O Alonzo-Proulx, Peressoti K,
Al-Mayah A, Brock K. The Myth of the 50%250% breast. Med Phys 2009;36:54375443.
1.14E201
250

951
2.75E103
9.81E102
4.51E102
2.44E102
1.47E102
9.49E101
6.44E101
4.57E101
3.37E101
2.02E101
1.26E101
8.49E100
6.02E100
4.44E100
2.37E100
1.43E100
9.63E201
7.00E201
10
12
14
16
18
20
25
30
35
40
7.17E100
1.04E101
1.60E101
2.62E101
1.77E101
9.20E100
1.11E101
1.82E101
2.29E101
1.82E101
8.87E100
1.45E101
3.54E101
2.66E101
4.39E101
3.32E101
4.87E101
6.97E101
1.06E102
1.71E102
2.25E102
3.04E102
4.33E102
6.44E102
5.20E102
3.80E102
7.96E102
2.11E103
5 4.51
CsI
6.39E101
6.01E101
8.54E101
8.16E101
1.32E102
1.73E102
2.37E102
3.39E102
4.96E102
2.93E102
4.37E102
9.20E102
2.48E103
5 4.56
BaFBr
4.80E101
8.66E101
1.22E102
2.64E101
4.34E101
5.81E101
8.08E101
1.16E102
1.78E102
2.88E102
5.22E102
1.11E103
3.09E103
5 4.79
5 2.33
ENERGY (keV)
Se (ELEMENTAL)
Si (ELEMENTAL)
5.72E100
8.22E100
1.25E101
2.02E101
3.68E101
4.86E101
6.64E101
9.47E101
1.43E102
2.27E102
2.96E102
3.44E102
1.39E102
2.12E102
3.35E102
5.90E102
1.22E103
2.88E103
5 7.34
Gd2O2S
8.40E100
1.20E101
1.82E101
2.93E101
5.26E101
6.93E101
6.70E101
9.29E101
1.38E102
1.46E102
9.01E101
1.23E102
1.74E102
2.64E102
4.14E102
7.28E102
1.49E103
2.37E103
5 7.57
YTaO4
TABLE D-3 MASS ENERGY ATTENUATION COEFFICIENTS IN cm2/g (DENSITY () IN g/cm3)
D.3 Mass Energy Attenuation Coefficients for Selected Detector Compounds
7.06E100
1.01E101
1.52E101
2.44E101
4.39E101
5.78E101
7.87E101
1.11E102
1.43E102
7.39E101
9.81E101
1.34E102
1.88E102
2.85E102
4.46E102
6.47E102
1.31E103
2.76E103
5 6.12
CaWO4
1.33E101
1.91E101
2.83E101
1.77E101
3.27E101
4.31E101
5.88E101
8.36E101
5.51E101
8.99E101
1.19E102
1.63E102
2.33E102
3.52E102
5.62E102
9.89E102
8.14E102
2.24E103
5 6.47
AgBr
952
1.92E201
1.84E201
1.72E201
1.63E201
1.56E201
95
100
110
120
130
1.33E201
2.00E201
90
1.37E201
2.11E201
85
180
2.23E201
80
170
2.40E201
75
1.40E201
2.59E201
70
160
2.87E201
65
1.50E201
3.22E201
60
1.45E201
3.70E201
55
140
4.39E201
50
150
5.41E201
45
2.02E201
2.21E201
2.43E201
2.72E201
3.08E201
3.59E201
4.23E201
5.07E201
6.31E201
7.22E201
8.12E201
9.37E201
1.09E100
1.29E100
1.54E100
1.90E100
2.35E100
2.98E100
3.88E100
5.18E100
3.95E201
4.51E201
5.10E201
5.91E201
6.93E201
8.38E201
1.01E100
1.25E100
1.60E100
1.85E100
2.10E100
2.47E100
2.86E100
3.39E100
4.07E100
5.01E100
6.17E100
7.72E100
9.91E100
1.30E101
4.80E201
5.54E201
6.28E201
7.33E201
8.65E201
1.05E100
1.28E100
1.59E100
2.04E100
2.37E100
2.69E100
3.18E100
3.68E100
4.37E100
5.25E100
6.46E100
7.96E100
1.00E101
1.29E101
1.69E101
6.15E201
7.11E201
8.09E201
9.44E201
1.12E100
1.36E100
1.65E100
2.04E100
2.61E100
3.02E100
3.44E100
4.06E100
4.69E100
5.55E100
6.59E100
8.05E100
9.83E100
1.23E101
3.07E100
4.15E100
6.26E201
7.22E201
8.21E201
9.57E201
1.13E100
1.37E100
1.68E100
2.06E100
2.61E100
3.00E100
3.40E100
3.97E100
4.57E100
5.40E100
6.42E100
2.30E100
2.84E100
3.57E100
4.61E100
6.11E100
6.98E201
8.07E201
9.18E201
1.07E100
1.26E100
1.52E100
1.87E100
2.28E100
2.90E100
3.33E100
3.77E100
4.41E100
5.07E100
5.98E100
7.06E100
1.98E100
2.44E100
3.05E100
3.92E100
5.17E100
(Continued)
3.04E201
3.43E201
3.84E201
4.40E201
5.12E201
6.13E201
7.36E201
9.05E201
1.15E100
1.33E100
1.51E100
1.77E100
2.05E100
2.44E100
2.93E100
3.62E100
4.50E100
5.64E100
7.29E100
9.68E100

953
5 4.79
5 2.33
1.16E201
1.08E201
1.01E201
9.59E202
9.14E202
8.73E202
8.38E202
8.07E202
7.78E202
7.52E202
7.29E202
7.07E202
ENERGY (keV)
250
300
350
400
450
500
550
600
650
700
750
800
6.64E202
6.94E202
7.23E202
7.62E202
8.09E202
8.61E202
9.30E202
1.02E201
1.15E201
1.33E201
1.63E201
2.14E201
5 4.56
BaFBr
6.75E202
7.08E202
7.40E202
7.84E202
8.37E202
8.97E202
9.78E202
1.09E201
1.24E201
1.46E201
1.84E201
2.48E201
5 4.51
CsI
7.21E202
7.60E202
7.99E202
8.51E202
9.16E202
9.91E202
1.09E201
1.23E201
1.42E201
1.72E201
2.21E201
3.06E201
5 7.34
Gd2O2S
7.40E202
7.81E202
8.22E202
8.78E202
9.45E202
1.02E201
1.13E201
1.27E201
1.47E201
1.78E201
2.27E201
3.14E201
5 7.57
YTaO4
7.69E202
8.13E202
8.58E202
9.19E202
9.93E202
1.08E201
1.19E201
1.35E201
1.58E201
1.92E201
2.48E201
3.46E201
5 6.12
CaWO4
6.62E202
6.88E202
7.14E202
7.48E202
7.87E202
8.31E202
8.88E202
9.61E202
1.06E201
1.19E201
1.41E201
1.77E201
5 6.47
AgBr
Source of these coefficients is described in: Boone JM, Chavez AE. Comparison of x2ray cross sections for diagnostic and therapeutic medical physics. Med Phys 1996;23:19972005.
6.30E202
6.52E202
6.75E202
7.02E202
7.33E202
7.68E202
8.11E202
8.63E202
9.27E202
1.02E201
1.14E201
1.35E201
Se (ELEMENTAL)
Si (ELEMENTAL)
TABLE D-3 MASS ENERGY ATTENUATION COEFFICIENTS IN cm2/g (DENSITY () IN g/cm3) (Continued)
954
APPENDIX
Effective Doses, Organ Doses,

and Fetal Doses from Medical
Imaging Procedures
Estimates of effective dose and organ doses for a specific diagnostic procedure extend
over a range of values, and are dependent on many parameters such as image quality
(signal to noise and contrast to noise ratios), patient size, x-ray acquisition techniques,
and the application of dose reduction technologies. Methods to reduce radiation dose
include utilization of higher quantum detection efficiency digital radiographic detectors and application of image processing algorithms to reduce noise. In CT, they
include implementation of automatic tube current modulation as a function of tube
angle and patient attenuation, and deployment of statistical iterative reconstruction
techniques. As technology advances and improves, a trend towards lower radiation
dose should occur, which for many procedures will result in lower effective doses
than the values listed in these tables. The numbers of days of typical background
radiation equal to the average effective dose of the examination are provided to help
to place the magnitude of the exposure into perspective.
Table E-1 lists typical adult effective doses for various diagnostic radiology procedures. Table E-2 provides specific information for interventional examinations, Table
E-3 lists dose information for common CT procedures, and Table E-4 presents information on dose from dental imaging procedures.
Table E-5 provides information on organ doses based on typical techniques for
radiography and CT procedures. Table E-6 lists organ doses determined from direct
measurements of a 6-year-old pediatric anthropomorphic phantom for routine
abdominal and chest CT examination techniques at seven different sites in Japan,
along with effective dose. Table E-7 lists effective dose estimates of various pediatric
examinations including the common chest radiograph as well as CT of the head and
abdomen as a function of age, from neonate to 15 years old. Table E-8 lists the conceptus dose for various CT, radiography and fluoroscopy imaging procedures.
Table E-1ADULT EFFECTIVE DOSES FOR VARIOUS DIAGNOSTIC RADIOLOGY
PROCEDURES (2008)
EXAMINATION
AVERAGE EFFECTIVE
DOSE (mSv)
VALUES REPORTED
IN LITERATURE (mSv)
DAYS OF EQUIVALENT
BACKGROUND RADIATION
Skull
0.1
0.030.22
12
Cervical spine
0.2
0.070.3
24
Thoracic spine
0.61.4
118
Lumbar spine
1.5
0.51.8
177
Posteroanterior and
lateral study of chest
0.1
0.050.24
12
(Continued)
955
Table E-1ADULT EFFECTIVE DOSES FOR VARIOUS DIAGNOSTIC RADIOLOGY
PROCEDURES (2008) (Continued)

AVERAGE EFFECTIVE
DOSE (mSv)
EXAMINATION
VALUES REPORTED
IN LITERATURE (mSv)
DAYS OF EQUIVALENT
Posteroanterior study
of chest
0.02
0.0070.050
Two view digital

mammography
0.4
0.100.60
47
Abdomen
0.7
0.041.1
82
Pelvis
0.6
0.21.2
73
Hip
0.7
0.182.71
82
Shoulder
0.01
Knee
0.005
Other extremities
0.001
0.00020.1
<1
Dual x-ray absorptiometry (without CT)
0.001
0.0010.035
<1
Dual x-ray absorptiometry (with CT)
0.04
0.0030.06
Intravenous
urography
0.73.7
350
Upper gastrointestinal series
1.512
710
Small-bowel series
3.07.8
590
Barium enema
2.018.0
940
Source: Mettler FA, et al.: Effective doses in radiology and diagnostic nuclear medicine: a catalog. Radiology
2008;248:254263.
TABLE E-2ADULT EFFECTIVE DOSES FOR VARIOUS INTERVENTIONAL
RADIOLOGY PROCEDURES (2008)

EXAMINATION
AVERAGE EFFECTIVE VALUES REPORTED

DAYS OF EQUIVALENT
DOSE (mSv)a
IN LITERATURE (mSv) BACKGROUND RADIATION
Head and/or neck

angiography
0.819.6
590
Coronary angiography
(diagnostic)
2.015.8
820
Coronary percutaneous
transluminal angioplasty,
stent placement, or
radiofrequency ablation
15
6.957
1,770
Thoracic angiography of
pulmonary artery or aorta
4.19.0
590
Endoscopic retrograde
cholangiopancreatography
2.08.0
470
Abdominal angiography or
aortography
12
4.048.0
1,410
Transjugular intrahepatic
portosystemic shunt
placement
70
20180
8,240
Pelvic vein embolization
60
4478
7,060
2008;248:254263.
Appendix E Organ Dose and Fetal Dose for Medical Imaging Procedures
957
TABLE E-3ADULT EFFECTIVE DOSES FOR VARIOUS CT PROCEDURES (2008)

AVERAGE EFFECTIVE
DOSE (mSv)
EXAMINATION
Head
RANGE REPORTED IN
LITERATURE (mSv)
DAYS OF EQUIVALENT
0.94.0
240
Neck
350
Chest
4.018.0
820
Chest for pulmonary

embolism
15
1340
1,770
Abdomen
3.525
940
Pelvis
3.310
710
Three-phase liver study
15
1,880
Spine
1.510
710
Coronary angiography
16
5.032
1,880
Calcium scoring
1.012
350
Virtual colonoscopy
10
4.013.2
1,180
Extremities
<0.025
<3
2008;248:254263.
TABLE E-4ADULT EFFECTIVE DOSE FOR VARIOUS DENTAL RADIOLOGY
PROCEDURES (2008)
AVERAGE EFFECTIVE
DOSE (mSv)
VALUES REPORTED
IN LITERATURE (mSv)
0.005
0.00020.010
Panoramic radiography
0.01
0.0070.090
Dental CT
0.2
20
EXAMINATION
Intraoral radiography
DAYS OF EQUIVALENT
2008;248:254263.
TABLE E-5TYPICAL ORGAN-SPECIFIC RADIATION DOSES RESULTING FROM
VARIOUS RADIOLOGY PROCEDURES

EXAMINATION
ORGAN
PA chest radiography
Lung
0.01
Mammography
Breast
3.5
CT chest
Breast
21.4
CT coronary angiography
Breast
51.0
Abdominal radiography
Stomach
CT abdomen
Stomach
Colon
10.0
4.0
Barium enema
Colon
15.0
Source: Davies HE, Wathen CG, Gleeson FV. BMJ 2011;342:d947.
ORGAN SPECIFIC RADIATION DOSE (mGy)
0.25
958
TABLE E-6ORGAN DOSE AVERAGES FOR A 6-YEAR-
OLD PEDIATRIC ANTHROPOMORPHIC

PHANTOM USING ROUTINE TECHNIQUES
AT SEVEN CT SCANNER SITES
ORGAN DOSE (mGy) 6
TISSUE OR ORGAN
PEDIATRIC ABDOMEN
PEDIATRIC CHEST
Thyroid gland
0.3 6 0.2
10.5 6 6.6
Lung
4.2 6 2.1
9.1 6 4.2
Breast
2.3 6 1.8
8.4 6 4.7
Esophagus
4.2 6 2.0
9.0 6 4.4
Liver
8.8 6 3.5
8.0 6 3.7
Stomach
9.5 6 3.9
4.7 6 2.8
Kidneys
9.0 6 3.5
4.3 62.2
Colon
9.4 6 3.5
0.6 6 0.3
Ovary
9.0 6 3.1
0.1 6 0.1
Bladder
9.1 6 3.3
0.1 6 0.0
Testis
7.8 6 3.7
0.1 6 0.0
Bone surface
8.1 6 2.8
8.5 6 3.8
Source: Fujii K, Aoyama T, Koyama S, et al. Comparative evaluation of organ

and effective doses for paediatric patients with those for adults in chest and
abdominal CT examinations. Br J Radiol 2007;80:657667.
, standard deivation
Table E-7EFFECTIVE DOSE ESTIMATES FROM PEDIATRIC DIAGNOSTIC
EXAMS AS A FUNCTION OF AGE

EXAMINATION
Chest (PA and Lat)
EFFECTIVE DOSE (mSv)
0.06
DAYS OF EQUIVALENT
CT Heada
neonate
4.2
490
1y
3.6
420
5y
2.4
280
10 y
2.0
240
15 y
1.4
160
neonate
13.1
1,540
1y
11.1
1,310
5y
8.4
990
10 y
8.9
1,050
15 y
5.9
690
CT Abdomena
Data from Thomas KE, Wang B. Age specific effective doses for pediatric MSCT examinations at a
large childrens hospital using DLP conversion coefficients: a simple estimation method. Pediatr Radiol
2008;38(6):645656.
Appendix E Organ Dose and Fetal Dose for Medical Imaging Procedures
959
TABLE E-8ESTIMATED CONCEPTUS DOSES FROM COMMON
RADIOGRAPHIC, FLUOROSCOPIC, AND CT EXAMINATIONS

ESTIMATED CONCEPTUS DOSES FROM SINGLE CT ACQUISITION
EXAMINATION
DOSE LEVEL
TYPICAL CONCEPTUS DOSE (mGy)
Extra-abdominal
Head CT
Standard
Routine
Standard
0.2
Pulmonary embolus
Standard
0.2
CT angiography of coronary arteries
Standard
0.1
Chest CT
Abdominal
Abdomen, routine
Standard
Abdomen/pelvis, routine
Standard
25
CT angiography of aorta (chest

through pelvis)
Standard
34
Abdomen/pelvis, stone protocola
Reduced
10
Anatomic coverage is the same as for routine abdominopelvic CT, but the tube current is decreased and the
pitch is increased because standard image quality is not necessary for detection of high-contrast stones.
ESTIMATED CONCEPTUS DOSES FROM RADIOGRAPHIC AND FLUOROSCOPIC

EXAMINATIONS
TYPICAL CONCEPTUS
DOSE (mGy)
EXAMINATION
Cervical spine (AP, lat)
<0.001
Extremities
<0.001
Chest (PA, lat)
0.002
Thoracic spine (AP, lat)
0.003
Abdomen (AP)
21-cm patient thickness
33-cm patient thickness
Lumbar spine (AP, lat)
Limited IVPa
6
b
Small-bowel study
Double-contrast barium enema studyc
AP, anteroposterior projection, lat, lateral projection, PA, posteroanterior projection.

a
Limited IVP is assumed to include four abdominopelvic images. A patient thickness of 21 cm is assumed.
b
A small-bowel study is assumed to include a 6-min fluoroscopic examination with the acquisition of 20 digital
spot images.
c
A double-contrast barium enema study is assumed to include a 4-min fluoroscopic examination with the
acquisition of 12 digital spot images.
Source: McCollough CH, Schueler BA, Atwell TD, et al. Radiation exposure and pregnancy: when should we be
concerned? Radiographics 2007;27:909917.
APPENDIX
Radiopharmaceutical
Characteristics and Dosimetry
960
METHOD
OF ADMb
ORL
IV
IV
Co-57 Vitamin B12 (DNI)
Cr-51 Sodium Chromate

RBCs (DNI)
F-18 Fluorodeoxyglucose
(DI)
(1)
RADIOPHARMACEUTICAL
(DI/DNI/T)a
4560 min
20 mL blood sample
@ 24 h and every 23
d for 30 d
NA
In vitro measurement
of radioactivity in
urine
NA
DELAY BEFORE
IMAGING
Glucose analogue,
increased uptake correlates
with increased glucose
metabolism. Crosses bloodbrain barrier, taken up by
cells where it is trapped due
to phosphorylation of FDG
6-phosphate.
Cr6+ attaches to hemoglobin

and reduced to Cr3+. Cr-RBC is
a blood pool marker.
Complexed to intrinsic factor

and absorbed by distal ileum.
METHOD OF LOCALIZATION
COMMONLY UTILIZED RADIOPHARMACEUTICALS
Major clinical application

in oncology imaging. Used
to assess recurrence and to
differentiate between recurring
tumor and necrotic tissue.
Also used for the localization
of epileptogenic focus and
evaluation of dementias.
Cardiac applications in
determining metabolic activity
and viability of myocardium.
Most commonly used for red

cell mass and survival. Also used
for splenic sequestration studies.
Non-imaging in vitro assay via
counting serial blood samples.
Diagnosis of pernicious anemia

and intestinal absorption
deficiencies.
CLINICAL UTILITY
(Continued)
PET Radiopharmaceutical.
511-keV annihilation photons
require substantial shielding
of source vials and care to
prevent cross-talk between
other imaging devices.
RBC labeling efficiency

80%90%. Free Cr3+ will
not tag RBCs and is rapidly
eliminated in the urine.
Splenic sequestration study.
~111 MBq (~3 mCi) of heat
denatured RBC.
Wait 24 h following more

than 1 mg IV/IM vitamin B12.
Have patient fast 8 h before
test.
COMMENTS
APPENDIX F-1METHOD OF ADMINISTRATION, LOCALIZATION, CLINICAL UTILITY, AND OTHER CHARACTERISTICS OF

Appendix F Radiopharmaceutical Characteristics and Dosimetry
961
METHOD
OF ADMb
IV
IV
IV
F-18 Sodium fluoride (DI)
Ga-67 Citrate (DI)
I-123 ioflupane also

known as DaTscan (DI)
(1)
RADIOPHARMACEUTICAL
(DI/DNI/T)a
36 h
Typically 2472 h
range 6120 h
12 h
DELAY BEFORE
IMAGING
CLINICAL UTILITY
Bone scan. Uptake of F-18

reects blood ow and
bone remodeling. The CT
component of PET/CT systems
have signicantly improved
the specicity, allowing for
morphologic characterization of
the functional lesions and more
accurate differentiation between
benign lesions and metastases.
Ioflupane binds to dopamine

transporter in the brain.
Brain SPECT for dopamine

transporter allows visualization
of brain striata to evaluate
patients with suspected
parkinsonian syndromes. Helps
differentiate essential tremor
from tremor due to Parkinsons.
Exact mechanism unknown.

Tumor, abscess imaging;
Accumulates in lysosomes and Hodgkins disease; lymphoma,
is bound to transferrin blood. and acute inflammatory lesions.
Interstitial lung diseases and
fever of unknown origin.
Fluorine ions accumulate in

the skeleton with deposition
being greater around active
areas (e.g., arthritis, trauma,
malignancies).
COMMONLY UTILIZED RADIOPHARMACEUTICALS (Continued)
Supersaturated potassium
iodide (SSKI) is administered
to block the thyroid. Inject
slowly over not less than
1520 s
Ga-67 uptake influenced

by vascularity, increased
permeability of tumor cells,
rapid proliferation, and low
pH. Used in lieu of WBC in
patients with leukemias or
white cell deficiencies. Ga-67
secreted in large intestine;
bowel activity interferes with
abdominal imaging.
Cleared rapidly in urine.

Patient should be well
hydrated prior to dosing.
Patient should void 30 min
after injection and again
immediately prior to imaging
to reduce background.
COMMENTS
962
46 h
ORL
IV
ORL
I-123 Sodium Iodide (DI

& DNI)
I-125 Albumin (DNI)
I-131 Sodium Iodide (DI,

DNI, T)
24 h
N/A
~20 mL blood withdrawn ~10 min after
dose administration
1824 h
I-123
IV
Metaiodobenzylguanidine
MIBG (DI)
Detection of primary or
metastatic pheochromocytoma
or neuroblastoma.
Blood and plasma volume

determinations.
Rapidly absorbed from GI

370 kBq (10 mCi) used for
tract, trapped and organically uptake studies (DNI).
bound by thyroid.
74185 MBq (25 mCi) for
diagnostic thyroid carcinoma
metastatic survey.
185555 MBq (515 mCi) or
hyperthyroidism therapy.
1.17.4 GBq (30200 mCi) for
thyroid carcinoma therapy.
Blood pool marker.
Rapidly absorbed from GI

Evaluation of thyroid function
tract. Trapped and organically and/or morphology. Uptake
bound by thyroid.
determined by Nal (TI) thyroid
probe.
Uptake and accumulation

by the norepinephrine
transporter in adrenergically
innervated tissues.
(Continued)
Liquid doses should be

manipulated in 100% exhaust
fume hood. Iodine is volatile
at low pH. 90% of local
radiation dose is from beta;
10% gamma.
Dilution principle and

hematocrit used to determine
blood and plasma volume.
Dose range 3.7 MBq (100 mCi)

(uptake study only) to
14.8 MBq (400 mCi) (uptake
and imaging). Up to 2 mCi for
metastatic disease imaging.
Administered as a capsule.
SSKI is administered to
block the thyroid. Weight
adjust for pediatric use, see
AppendixF-3B. SPECT may
be performed after planar
imaging. False negatives seen
in patients with impaired
renal function. MIBG is not
cleared by dialysis. I-131 MIBG
for therapy is in clinical trials.

963
METHOD
OF ADMb
IV
IV
IV
I-131 Tositumomab also

known as Bexxar (T)
In-111 Capromab
Pendetide also known as
ProstaScint (DI)
In -111 Pentetreotide also

known as Octreoscan (DI)
(1)
RADIOPHARMACEUTICAL
(DI/DNI/T)a
2448 h
72120 h
Whole-body counts
are obtained on day
1 and 3 following
imaging dose
~74185 MBq
(~25 mCi) to
determine residence
time and calculate
therapeutic dose
DELAY BEFORE
IMAGING
Pentetreotide is a DTPA
conjugate of octreotide,
a long acting analog of
the human hormone
somatostatin. Pentetreotide
binds to somatostatin
receptors on the cell surfaces
throughout the body.
This agent is a monoclonal

antibody to a membrane
specific antigen expressed in
many primary and metastatic
prostate cancer lesions.
This agent is a radiolabeled

murine IgG2a monoclonal
antibody directed against the
CD20 antigen on B cells.
Localization of primary and

metastatic neuroendocrine
tumors bearing the
somatostatin receptor.
Indicated for patients with

biopsy proven prostate cancer
that is thought to be clinically
localized following standard
diagnostic imaging tests. Also
indicated in postprostatectomy
patients with high clinical
suspicion of occult metastatic
disease.
Radioimmunotherapic
monoclonal antibody for the
treatment of chemotherapyrefractory low-grade or
transformed low-grade
Non-Hodgkins Lymphoma
CLINICAL UTILITY
Incubate In-111 Pentetreotide

at 25C for at lease 30 min.
Do not administer in TPN
admixtures or lines as complex
glycosyl octreotide conjugates
may form. Patients should be
well hydrated prior to dose
administration. Clinical trials
underway for use of this as a
therapeutic agent with Y-90.
High false positive rate.

Best results when imaged
on a SPECT/CT. Patient
management should not
be based on ProstaScint
results without confirmatory
studies. May induce human
anti-mouse antibodies which
may interfere with some
immunoassays (e.g., PSA and
dioxin).
Cytotoxic effect is the result of

both direct antitumor effects
of the antibody and the
radiation dose delivered by the
I-131. 90% of local radiation
dose is from beta; 10%
gamma. SSKI is administered
to block the thyroid. Can
typically be administered on
an outpatient basis.
COMMENTS
964
NA
NA
IC
P-32 Chromic Phosphate

Colloid (T)
P-32 Sodium Phosphate (T) IV
None
INH
Kr-81m Krypton Gas (DI)
424 h
IV
In-111 White Blood Cells

(DI)
Concentration in blood cell

precursors.
Intraperitoneal (IP) or
intrapleural (IPL) instillation.
No equilibrium or washout
phase, thus Kr-81m
distribution reflects only
regional ventilation.
In-111 oxyquinoline is a
neutral lipid-soluble complex
that can penetrate cell
membranes where the In-111
translocates to cytoplasmic
components.
Treatment of polycythemia
vera (PCV) most common; also
myelocytic leukemia, chronic
lymphocytic leukemia.
Instilled directly into the body

cavity containing a malignant
effusion. Dose range (IP)
370740 MBq (1020 mCi) (IPL)
222444 MBq (612 mCi).
Lung ventilation studies;

however, availability of Rb-81/
Kr- 81m generator is limited due
to short parent T1/2 (4.6 h). Used
infrequently due to expense for
the generator.
Detection of occult
inflammatory lesions not
visualized by other modalities.
Confirmation of active infection
in a known area. Careful
isolation and labeling of WBCs
is essential to maintain cell
viability labeling efficiency
~80%. Better for deep abscesses
or vertebral osteomyelitis than
Tc-99m WBC.
(Continued)
Clear colorless solution,

typical dose range (PCV)
37296 MBq (18 mCi) Do
not use sequentially with
chemotherapy. WBC should be
5,000/mm3; platelets should
be 150,000 mm3.
Bluish green colloidal

suspension. Never given IV.
Do not use in the presence of
ulcerative tumors.
Due to ultrashort half life

(13 s) only wash-in images can
be obtained, however, repeat
studies and multiple views can
be obtained; radiation dose
and hazard is minimal.
In-111 WBC prepared in vitro

with In-111 Oxyquinoline
(Indium Oxine) and isolated
plasma free suspension of
autologous WBCs. 30% dose
in spleen, 30% liver, 5%
lung (clears 4 h). 20% in
circulation after injection.
In-111 WBC has a longer T1/2
and holds the label longer than
Tc-99m WBC.

965
METHOD
OF ADMb
IV
IV
Over 1
to 2 min
period
IV
Over a
12 min
period
Rb-82 also known as

CardioGen-82 (DI)
Sm-153 EDTMP also

known as Quadramet (T)
Sr-89 Chloride also known

as Metastron (T)
(1)
RADIOPHARMACEUTICAL
(DI/DNI/T)a
N/A
103-keV photon can

be imaged however
this is not essential
Infused into patient

while on camera.
Begin imaging
1060 s after infusion
when count rate
reaches 550 kcpm.
DELAY BEFORE
IMAGING
Calcium analog selectively

localizing in bone mineral.
Preferential absorption occurs
in sites of active osteogenesis
(i.e., primary and metastatic
bone tumors).
This therapeutic agent

is formulated as a
tetraphosphate chelate
(EDTMP: ethylenediaminetetra
methylenephosphonic acid)
with an affinity for areas of
high bone turnover associated
with the hydroxyapatite (i.e.,
primary and metastatic bone
tumors).
Accumulates in viable
myocardium analogous
to potassium. Myocardial
extraction fraction is 60% at
rest and 30% at peak stress.
Therapy only. Indicated for

the palliative relief of bone
pain in patients with painful
skeletal metastasis. Pure beta
so radiation safety concerns
for caregivers are less than for
Sm-153.
Therapy only. Indicated for the

palliative relief of bone pain in
patients with painful skeletal
metastases. Beta and gamma.
Gamma allows imaging if
desired.
Myocardial perfusion agent

that is useful in distinguishing
normal from abnormal
myocardium in patients with
suspected myocardial infarction.
Short half life allows rest and
stress to be performed in less
than 1 h.
CLINICAL UTILITY
Bone marrow toxicity is

expected. Requires minimum
WBC/platelet count.
Peripheral blood cell counts
should be monitored at least
once every other week. Pain
relief expected 721 days
following dose. Due to longer
half-life, Metastron is not
recommended in patients with
very short life expectancy.
The patient should be well

hydrated. Bone marrow
toxicity is expected. Requires
assessment of WBC/platelet
count prior to use. Peripheral
blood cell counts should
be monitored at least once
every other week. Pain relief
expected 721 days following
dose.
Better imaging agent for large

patients (i.e., less attenuation).
No ability to perform exercise
stress tests. Must have a
generator on site so high
patient volume is essential to
be cost effective. Sr-82/Sr-85
breakthrough was recently
the subject of an FDA safety
advisory.
COMMENTS
966
IV
IV
IV
Tc-99m Disofenin
also known as HIDA
(iminodiacetic acid) (DI)
Tc-99m DMSA
(dimercaptosuccinic acid)
also known as Succimer
(DI)
Tc-99m Exametazime also

known as Ceretec and
HMPAO (DI)
Ceretec is used to radiolabel

leukocytes as an adjunct in the
localization of intra-abdominal
infection and inflammatory
bowel disease. Good for
superficial lesions such as
diabetic ulcers. Also used in
detection of altered regional
cerebral perfusion in stroke,
Alzheimers disease, seizures, etc.
Rapidly cleared from blood.

Lipophilic complex, brain
uptake 4% of injected dose
occurs 1 min after injection.
~30300 min.
When used for
inflammatory bowel
or appendicitis,
imaging must start
immediately.
Hepatobiliary imaging agent,

hepatic duct and gallbladder
visualization occurs 20 min
after administration as liver
activity decreases. The patient
should be fasting 4 h. Delayed
images are obtained @ 4 h for
acute/chronic cholecystitis.
Evaluation of renal parenchymal

After IV administration
disorders.
Tc-99m succimer is bound to
plasma proteins. Cleared from
plasma with a T1/2 of about
60 min. and concentrates in
the renal cortex.
Excreted through the

hepatobiliary tract into
the intestine. Clearance
mechanism: hepatocyte
uptake, binding and storage.
Excretion into canaliculi via
active transport.
12 h
5 min serial images

every 5 min for 30
min; and @ 40
& 60 min.
(Continued)
Administer ASAP after

preparation. SPECT images
only. High photon flux
and fast uptake in sites
of infection permits early
imaging. Short half-life limits
delayed imaging.
Should be administered
within 10 min to 4 h following
preparation. Best cortical
imaging agent. Only renal
tracer for use with SPECT/CT.
Biliary system not well

visualized when serum
bilirubin values are over
8 mg/dL; do not use if over
10 mg/dL. Normal study:
visualized common duct and
intestinal tract by 30 min.
CCK routinely used for gall
bladder ejection fraction.

967
IV
Tc-99m Mertiatide also

known as MAG3 (DI)
None
24 h
IV
Tc-99m phosphonates
also known as Tc-99m
Methyenediphosphonate
(MDP) (DI)
DELAY BEFORE
IMAGING
None
METHOD
OF ADMb
Tc-99m Macro aggregated IV

albumin (MAA) (DI)
(1)
RADIOPHARMACEUTICAL
(DI/DNI/T)a
Reversible plasma protein

binding. Excreted by kidneys
via active tubular secretion
and glomerular filtration.
Specific affinity for areas of

altered osteogenesis uptake
related to osteogenic activity
and skeletal perfusion. 50%
bone 50% bladder in 24 h.
Renal imaging agent.

Diagnostic aid for assessment
of renal function; split function;
renal angiograms; renogram
curves, whole kidney and renal
cortex. Best tubular agent with
high extraction faction. MAG3
clearance provides estimate of
effective renal plasma flow.
Skeletal imaging agent.

Three-phase (flow, blood
pool, bone uptake) used to
distinguish between cellulitis and
osteomyelitis. Theory of bone
uptake: (1) hydroxyapatite crystal
binding; (2) collagen dependent
uptake in organic bone matrix.
CLINICAL UTILITY
Scintigraphic evaluation of
pulmonary perfusion. 0.6
million (range 0.21.2) particles
are injected with an adult dose.
Not be used in patients with
severe pulmonary hypertension
or history of hypersensitivity
to human serum albumin. Also
used to assess liver perfusion
before chemoembolization or
sphere therapy.
85% Tc-99m MAA

mechanically trapped in
arterioles and capillaries.
Particle size (1580 mm).
Distribution in lungs is
a function of regional
pulmonary blood flow.
90% of dose excreted in 3 h.

Preparation involves injection
of 2 mL of air in reaction
vial and 10-min incubation in
boiling water bath within 5
min after Tc-99m is added. Kits
are light sensitive therefore
must be stored in dark.
Normal increase in activity in

distal aspect of long bones
compared to diaphyses.
Higher target to background
than pyrophosphate; 85%
urinary excretion w/MDP
versus 65% w/pyrophosphate
in 24 h.
Mix reagent thoroughly

before withdrawing dose;
avoid drawing blood into
syringe or injecting through IV
tubing to prevent nonuniform
distribution of particles. Inject
patients supine.
COMMENTS
968
IV
IV
IV
IV
Tc-99m Bicisate also

known as Neurolite (DI)
Tc-99m Pentetate also

known as Tc-99m DTPA
(DI)
Tc-99m Pyrophosphate
(DI)
Tc-99m Red Blood Cells

(DI)
None
16 h bone imaging;
12 h myocardial
imaging.
None
3060 min
Skeletal imaging agent used

to demonstrate areas of hyper
or hypo osteogenesis and/or
osseous blood flow; cardiac
agent as an adjunct in diagnosis
of acute myocardial infarction.
Blood pool imaging including
cardiac first pass and gated
equilibrium imaging and
detection of sites of G.I.
bleeding. Heat damaged RBC are
used for diagnosis of splenosis
and accessory spleen. Also used
for diagnosis of hemangiomas.
Blood pool marker 25%

excreted in 24 h. 95% of
blood pool activity bound to
RBC @ 24 h.
Kidney imaging; assess renal

perfusion; estimate glomerular
filtration rate (GFR; 20%
filtered per pass). Historically
used as a brain imaging agent
to identity excessive vascularity
or altered blood-brain barrier.
SPECT imaging with Neurolite

used as an adjunct to CT and
MRI in the localization of stroke
in patients in whom stroke has
been diagnosed.
Specific affinity for areas

of altered osteogenesis
and injured/infarcted
myocardium. 50% (bone) and
50% (bladder) in 24 h.
Cleared by glomerular
filtration, little to no binding
to renal parenchyma.
This is a lipophilic complex

that crosses the blood-brain
barrier and intact cells by
passive diffusion. Cells
metabolize this agent to a
polar (i.e., less diffusible)
compound.
(Continued)
In vitro cell labeling with

kit containing lyophilized
tin chloride dihydrate, Na
citrate and dextrose. Blood
collection with 1921 gauge
needle to prevent hemolysis.
Use heparin or ACD for
anticoagulants, not EDTA
or oxalatet. Two syringe
method and gentle mixing
complete the radiolabeling;
administered within 30 min.
labeling efficiency 95%.
Pyrophosphate also has an

affinity for red blood cells.
Administered IV 30 min
before Tc-99m for in vivo RBC
cell labeling. 75% of activity
remains in blood pool.
5% bound to serum proteins,

which leads to GFR lower than
that determined by inulin
clearance. Can be used in
aerosol form for ventilation
side of VQ scan.
Localization depends upon

both perfusion and regional
uptake by brain cells.
Use caution in patients with
hepatic or renal impairment.

969
METHOD
OF ADMb
IV
IV
IV
Tc-99m Sestamibi also

known as Cardiolite (DI)
Tc-99m Sodium
Pertechnetate (DI)
Tc-99m Sulfur Colloid (DI)
(1)
RADIOPHARMACEUTICAL
(DI/DNI/T)a
Pertechnetate ion distributes

similarly to the iodide ion; it is
trapped but not organified by
the thyroid gland.
Rapidly cleared by
reticuloendothelial (RE)
system. ~85% colloidal
particles are phagocytized by
Kupffers cells of liver. ~10%
in RE system of spleen; ~5% in
bone marrow.
~20 min
myocardium in a manner
analogous to TI-201 chloride.
Major clearance pathway is
the hepatobiliary system.
None (Angio/
Venography).
Other applications
~0.51 h.
3060 min
DELAY BEFORE
IMAGING
Relative functional assessment

of liver, spleen, bone
marrow RE system; also
used for gastroesophageal
reflux imaging; esophageal
transit time following oral
administration. When used as
a marrow agent, correlates
with WBC imaging. Patency
evaluation of peritoneovenous
(LeVeen) shunt, administered in
shunt.
Primary agent for

radiopharmaceutical
preparations, thyroid imaging,
salivary gland imaging,
placental localization, detection
of vesicourethral reflux,
radionuclide angiography/
venography.
Diagnosis of myocardial
ischemia and infarct. Also
used for tumor imaging and
parathyroid imaging.
CLINICAL UTILITY
Preparation requires addition

of acidic solution; 5 min in
boiling water bath followed
by addition of a buffer to
bring pH to 67. Plasma
clearance T ~25 min.
Larger particles (~100 nm)
accumulate in liver and spleen;
small particles less than 20 nm
in bone marrow.
Eluate should not be used

greater than 12 h after
elution. Tc-99m produced as
daughter of Mo/Tc generator.
The patient should not

have fatty foods (e.g., milk)
following injection. Low-dose/
high-dose regime (typical 3:1
activity ratio) allows obtaining
same day rest/stress images.
COMMENTS
970
IV
IV
INH
Tc-99m Tetrofosmin also

known as Myoview (DI)
Tl-201 Thallous Chloride

(DI)
Xe-133 Xenon Gas (DI)
None
Injected at maximal
stress. Image
~1015 min later.
Redistribution image
~4 h after stress
image.
15-min stress 30-min

rest.
Identification and
semiquantitative assessment
of myocardial ischemia and
infarction. Delayed imaging
used for myocardial viability.
Useful in the delineation of

regions of reversible myocardial
ischemic in the presence
or absence of infarcted
myocardium.
Inhalation; Xe-133 freely

Pulmonary function. Gas is
exchanges between blood
administered by inhalation from
and tissue; concentrates more a closed respirator system.
in fat than other tissues.
myocardium analogous to
potassium.
Lipophilic cationic complex

whose uptake in the
myocardium reaches a
maximum of ~1.2% of the
injected dose @ 5 min post
injection.
(Continued)
Physiologically inert noble gas.

Most of the gas that enters
circulation returns to lungs
and is exhaled. Exhaled Xe-133
gas trapped (adsorbed) by
charcoal filter system.
Tc-99m and Rb-82-based

myocardial perfusion
agents are replacing Tl-201
due to superior imaging
characteristics and lower dose
to the patient. Advantage to
using Tl is the ability to do
dual isotope imaging when
used with Tc-99m agents.
Diagnostic efficacy for

ischemia and infarct are
similar to Tl-201. Background
clears faster than sestamibi
thus the delay before imaging
is reduced. Effective dose
to the patient is reduced by
almost 50% when using 99mTc
for both rest and stress studies
versus using Tl-201 for the rest
portion.

971
METHOD
OF ADMb
IV
IV
Y-90 Microspheres (T)
Y-90 Zevalin (T)
Pretherapy imaging
with In-111 Zevalin
to determine
residence time and
calculate therapeutic
dose
Imaging with
Tc99m-MAA prior to
therapy is critical to
assess any lung or GI
shunts.
DELAY BEFORE
IMAGING
This agent is a radiolabeled

murine IgG1 monoclonal
antibody directed against the
CD20 antigen on B cells.
Radioembolization, most
commonly of tumors in the
liver. Radiolabeled spheres
delivered via intra-hepatic
artery catheter to the liver
tumor and are trapped in
the microvasculature due to
arteriolar capillary blockade.
Radiopharmaceutical used for: DI, Diagnostic Imaging; DNI, Diagnostic Non-imaging; T, Therapy.
Method of Administration: IV, Intravenous; IC, Intracavitary; ORL, Oral; INH, Inhalation;
c
NA, Not applicable.
(1)
RADIOPHARMACEUTICAL
(DI/DNI/T)a
Radioimmunotherapeutic
monoclonal antibody for
the treatment of relapsed or
refractory low-grade, follicular
or transformed B cell nonHodgkins Lymphoma
Treatment of unresectable
hepatocellular carcinomas with
beads (glass or resin) tagged
with Y-90 that are delivered
via a catheter to the liver.
Bystander and/or radiation
crossfire effect on cells of the
microvasculature.
CLINICAL UTILITY
Cytotoxic effect is the result

of both direct antitumor
effects (apoptosis) by the
antibody and the radiation
dose delivered by the Y-90.
Bystander and/or radiation
crossfire effect on adjacent
cells. Can typically be
administered on an outpatient
basis.
Treatment is brachytherapy
carried out with a team
including interventional
radiologists and radiation
oncologists or nuclear
medicine physician. Two
options available, SirSpheres
and Theraspheres. Critical that
all shunts from the area of the
tumor be coiled or ballooned
off prior to treatment to
prevent beads from migrating
to the lungs or GI tract.
COMMENTS
972
MBq
0.037
7.4
185
740
185
185
185
14.8
0.74
3,145
RADIOPHARMACEUTICAL
Co-57 Vitamin B12 also

known as Schillings test
Cr-51 Sodium chromate

denatured RBCs
F-18 Sodium fluoride
F-18 Fluoro-deoxyglucose
Ga-67 Citrate
I-123 Loflupane also known

as DaTscan
I-123
(MIBG)
I-123 Sodium iodide (35%

uptake)
I-125 Albumin
I-131 Bexxar
85
0.02
0.4
20
0.2
0.001
mCi
TYPICAL ADULT
DOSE
Thyroid
Heart
Thyroid
Liver
Bladder
Bone surfaces
Bladder
Bladder
Spleen
Liver
ORGAN
RECEIVING
HIGHEST DOSE
8,523
0.51
66.6
12.4
9.81
116.55
96.2
40.7
41.44
0.89
mGy
852.3
0.05
6.66
1.24
0.98
11.66
9.62
4.07
4.14
0.09
rad
ORGAN DOSE
786.3
2,610
0.15
0.12
0.16
0.07
1.52
1.05
2.96
1.57
15.17
10.36
10.36
8.14
2.41
2.04
0.13
0.04
0.03
0.02
mGy
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
78.63
261
0.02
0.01
0.02
0.007
0.15
0.11
0.3
0.16
1.52
1.04
1.04
0.81
0.24
0.20
0.013
0.004
0.003
0.002
rad
GONADAL DOSE
0.25
2.22
2.41
3.94
18.5
14.06
2.96
0.1
mSv
0.03
0.222
0.241
0.394
1.85
1.406
0.5
0.296
0.01
rem
EFFECTIVE
DOSE
NA4
0.34
0.15
0.013
0.021
0.1
0.019
0.027
0.4
2.7
mSv/MBq
1.258
0.555
0.048
0.079
0.37
0.07
0.1
1.48
9.99
rem/mCi
EFFECTIVE DOSE PER

ADMINISTERED
ACTIVITY
(Continued)
SOURCE
AND ITS DOSE; GONADAL DOSE; EFFECTIVE DOSE AND EFFECTIVE DOSE PER ADMINISTERED ACTIVITY FOR
SOME COMMONLY USED DIAGNOSTIC AND THERAPEUTIC RADIOPHARMACEUTICALS
APPENDIX F-2SUMMARY OF TYPICALLY ADMINISTERED ADULT DOSE; ORGAN RECEIVING THE HIGHEST RADIATION DOSE

973
MBq
3,700
222
19
370
148
1,480
2,590
148
185
185
RADIOPHARMACEUTICAL

uptake)
In-111 Octreotide also

known as Octreoscan
In-111 White blood cells
Kr-81m Krypton gas
P-32 Sodium phosphate
Rb-82
Sm-153 EDTMP also known

as Quadramet
Sr-89 Chloride also known

as Metastron
Tc-99m Disofenin
also known as HIDA
(iminodiacetic acid)
Tc-99m DMSA
70
40
10
0.5
100
mCi
TYPICAL ADULT
DOSE
Kidneys
Gall bladder
Bone surfaces
Bone surfaces
Kidney
Red marrow
Lung
Spleen
Spleen
Thyroid
ORGAN
RECEIVING
HIGHEST DOSE
33.3
20.35
2.52E+03
1.75E+04
13.79
1,628
0.08
101.75
126.54
2.66E+05
mGy
3.33
2.04
2.52E+02
1.75E+03
1.38
162.8
0.008
10.18
12.65
2.66E+04
rad
ORGAN DOSE
0.65
0.33
3.52
0.28
115.44
115.44
22.27
13.99
0.48
0.29
109.52
109.52
6.3E-05
6.3E-06
2.22
0.83
5.99
3.77
162.8
107.3
mGy
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
0.07
0.03
0.35
0.03
11.54
11.54
2.23
1.40
0.05
0.03
10.95
10.95
6.3E-06
6.3E-07
0.22
0.08
0.60
0.38
16.28
10.73
rad
GONADAL DOSE
1.63
3.15
2.55
0.01
10.92
11.99
mSv
0.16
0.32
0.26
0.001
1.09
1.20
rem
EFFECTIVE
DOSE
0.009
0.017
NA4
NA
1.72E-03
NA4
2.70E-05
0.59
0.054
NA4
mSv/MBq
0.033
0.063
6.36E-03
9.99E-05
2.183
0.200
rem/mCi
EFFECTIVE DOSE PER

ADMINISTERED
ACTIVITY
SOURCE
SOME COMMONLY USED DIAGNOSTIC AND THERAPEUTIC RADIOPHARMACEUTICALS (Continued)
974
740
148
740
370
740
370
555
740
296
1,110
370

HMPAO
Tc-99m Macroaggregated
albumin (MAA)
Tc-99m phosphonates also

known as Tc-99m MDP

known as MAG3
Tc-99m Bicisate also known

as ECD and Neurolite

Tc-99m Red blood cells
Tc-99 Sestamibi also known

as Cardiolite (rest)

as Cardiolite (stress)
Tc-99m Sodium
pertechnetate
10
30
20
15
10
20
10
20
20
Upper large
intestine
Gall bladder
Gall bladder
Heart
Bone surfaces
Bladder
Bladder
Bladder
Bone surfaces
Lungs
Kidneys
21.09
36.63
11.54
17.02
34.97
22.94
37.0
40.7
46.62
9.77
25.16
2.11
3.66
1.54
1.70
3.50
2.30
3.70
4.07
4.66
0.98
2.52
3.7
1.04
8.99
4.11
2.69
1.12
2.74
1.70
2.0
1.33
1.55
1.07
5.85
2.0
2.0
1.37
2.66
1.78
0.27
0.16
4.88
1.78
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
0.37
0.10
0.90
0.41
0.27
0.11
0.27
0.17
0.2
0.13
0.16
0.11
0.59
0.20
0.20
0.14
0.27
0.18
0.03
0.02
0.49
0.18
4.81
8.77
2.66
5.18
3.16
1.81
5.70
2.59
4.22
1.63
6.88
0.48
0.88
0.27
0.52
0.32
0.18
0.57
0.26
0.42
0.16
0.69
0.013
0.008
0.009
0.007
0.006
0.005
0.008
0.007
0.006
0.011
0.009
0.048
0.029
0.033
0.026
0.021
0.018
0.029
0.026
0.021
0.041
0.034
(Continued)

975
296
1,110
74
555
38,332
Tc-99m Sulfur colloid

known as Myoview (rest)

known as Myoview (stress)
Tl-201 Thallous chloride
Xe-133 Xenon gas

(rebreathing for 5 min)
Y-90 Zevalin
mCi
1,036
15
30
Spleen
Lungs
Kidneys
Gall bladder
Gall bladder
Spleen
ORGAN
RECEIVING
HIGHEST DOSE
360,320
0.61
35.52
29.97
7.99
13.88
mGy
36,032
0.06
3.55
3.00
0.80
1.39
rad
ORGAN DOSE
15,332
34,822
0.41
0.38
8.88
13.32
8.44
3.22
2.28
1.01
0.41
0.1
mGy
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
ov
ts
1,533
34,882
0.04
0.04
0.89
1.33
0.84
0.32
0.23
0.101
0.04
0.01
rad
GONADAL DOSE
0.44
10.36
7.66
2.04
1.74
mSv
0.04
1.04
0.77
0.20
0.17
rem
EFFECTIVE
DOSE
NA4
0.0008
0.140
0.007
0.007
0.010
mSv/MBq
0.003
0.518
0.026
0.026
0.035
rem/mCi
EFFECTIVE DOSE PER

ADMINISTERED
ACTIVITY
SOURCE
1. Annals of the International Commission on Radiological Protection Publication 106: Radiation Dose to Patients from Radiopharmaceuticals, Elsevier Science Inc., Tarrytown, NY,
2009. Note: Effective dose calculated utilizing weighting factors from ICRP Report 60.
3. Annals of the International Commission on Radiological Protection Publication 53: Radiation Dose to Patients from Radiopharmaceuticals, Pergamon Press, Elmsford, NY, 1988.
Note: The value provided in the Effective Dose column is the effective dose equivalent as defined on pages 1011 of the report and utilizing weighting factors from ICRP Report 26.
4. The effective dose is not a relevant quantity for therapeutic doses of radionuclides because it provides an estimate of effective stochastic risk (e.g., cancer and genetic detriment)
and is not relevant to effects at high tissue/organ doses such as deterministic (i.e., nonstochastic) risks.
5. Information supplied by the manufacturer.
6. Stabin Michael G. Proposed Revision To the Radiation Dosimetry of 82Rb. Health Physics. 99(6):811813, December 2010.
NA Not applicable
E, exponential (e.g., 2.66 E+055 = 2.66 105.)
MBq
185
RADIOPHARMACEUTICAL
TYPICAL ADULT
DOSE
SOME COMMONLY USED DIAGNOSTIC AND THERAPEUTIC RADIOPHARMACEUTICALS (Continued)
976
0.024
0.130
0.017
0.230
0.071
0.459
0.021
0.011
0.011
0.016
0.007
F-18 Fluorodeoxyglucose
I-123
(MIBG)

uptake)
In-111 Octreotide also

known as Octreoscan
Tc-99m Disofenin
also known as HIDA
(iminodiacetic acid)
Tc-99m DMSA

HMPAO
albumin (MAA)
Tc-99m phosphonates also

known as Tc-99m MDP
0.034
Ga-67 Citrate
mSv/MBq
RADIOPHARMACEUTICAL
0.026
0.059
0.041
0.041
0.078
1.700
0.263
0.851
0.063
0.481
0.089
0.126
rem/mCi
15 years old
0.011
0.023
0.017
0.015
0.029
0.703
0.100
0.350
0.026
0.200
0.037
0.052
mSv/MBq
0.041
0.085
0.063
0.056
0.107
2.600
0.370
1.295
0.096
0.740
0.137
0.192
rem/mCi
10 years old
0.014
0.034
0.027
0.021
0.045
1.054
0.160
0.740
0.037
0.330
0.056
0.086
mSv/MBq
0.052
0.126
0.100
0.078
0.167
3.900
0.592
2.738
0.137
1.221
0.207
0.318
rem/mCi
5 years old
EFFECTIVE DOSE
COMMONLY USED DIAGNOSTIC RADIOPHARMACEUTICALS
0.027
0.063
0.049
0.037
0.100
1.919
0.280
1.400
0.068
0.640
0.095
0.170
mSv/MBq
0.100
0.233
0.181
0.137
0.370
7.100
1.036
5.180
0.252
2.368
0.352
0.629
rem/mCi
1 year old
(Continued)
SOURCE
APPENDIX F-3AEFFECTIVE DOSE PER UNIT ACTIVITY ADMINISTERED TO 15;10;5 AND 1-YEAR-OLD PATIENTS FOR SOME

977
0.026
0.033
0.010
0.006
0.007
Tc-99m Bicisate also known

as ECD and Neurolite

0.033
0.740
0.010
0.017
0.012
0.009
0.009
0.200
0.001

as Cardiolite (stress)
Tc-99m Sodium
pertechnetate

known as Myoview (rest)

known as Myoview (stress)
Tl-201 Thallous chloride
Xe-133 Xenon gas (5 min)
0.002
0.560
0.013
0.013
0.018
0.026
0.016
0.018
0.014
0.011
0.008
0.015
0.012
mSv/MBq
0.006
2.072
0.048
0.048
0.067
0.096
0.059
0.067
0.052
0.041
0.030
0.056
0.044
rem/mCi
10 years old
0.003
0.790
0.021
0.021
0.028
0.042
0.023
0.028
0.021
0.014
0.009
0.022
0.012
mSv/MBq
0.010
2.923
0.078
0.078
0.104
0.155
0.085
0.104
0.078
0.052
0.033
0.081
0.044
rem/mCi
5 years old
0.005
1.300
0.039
0.039
0.050
0.079
0.045
0.053
0.039
0.027
0.016
0.040
0.022
mSv/MBq
0.020
4.810
0.144
0.144
0.185
0.292
0.167
0.196
0.144
0.100
0.059
0.148
0.081
rem/mCi
1 year old
SOURCE
2. Annals of the International Commission on Radiological Protection Publication 80: Radiation Dose to Patients from Radiopharmaceuticals, Elsevier Science Inc., Tarrytown, NY, 1999.
Note: Effective dose calculated utilizing weighting factors from ICRP Report 60.
3. Annals of the International Commission on Radiological Protection Publication 53: Radiation Dose to Patients from Radiopharmaceuticals, Pergamon Press, Elmsford, NY, 1988.
Note: The value provided in the Effective Dose column is the effective dose equivalent as defined on pages 1011 of the report and utilizing weighting factors from ICRP Report 26.
0.004
0.033
0.044
0.063
0.037
0.044
0.009
0.012

as Cardiolite (rest)
0.022
0.037
0.033
0.009

known as MAG3
rem/mCi
mSv/MBq
RADIOPHARMACEUTICAL
15 years old
EFFECTIVE DOSE
COMMONLY USED DIAGNOSTIC RADIOPHARMACEUTICALS (Continued)
APPENDIX F-3AEFFECTIVE DOSE PER UNIT ACTIVITY ADMINISTERED TO 15;10;5 AND 1-YEAR-OLD PATIENTS FOR SOME
978
9.3 MBq/kg (0.25 mCi/kg)
Tc-MDP
Tc-iminodiacetic acid
99m
Tc-macroaggregated
albumin
99m
If 99mTc used for ventilation, 2.59

MBq/kg (0.07 mCi/kg)
No 99mTc ventilation study, 1.11
MBq/kg (0.03 mCi/kg)
37 MBq (1.0 mCi)
99m
14.8 MBq (0.4 mCi)
18.5 MBq (0.5 mCi)
18.5 MBq (0.5 mCi)
Without flow study, 3.7 MBq/kg

(0.10 mCi/kg)
With flow study, 5.55 MBq/kg
(0.15 mCi/kg)
Tc-MAG3
37 MBq (1.0 mCi)
37 MBq (1.0 mCi)
37 MBq (1.0 mCi)
148 MBq (4 mCi)
370 MBq (10.0 mCi)
Maximum
MINIMUM ADMINISTERED administered
ACTIVITY
activity
Tc-dimercaptosuccinic acid 1.85 MBq/kg (0.05 mCi/kg)
F-FDG
99m
18
99m
Body, 3.75.2 MBq/kg (0.100.

14 mCi/kg)
Brain, 3.7 MBq/kg
(0.10 mCi/kg)
123
I-MIBG
Radiopharmaceutical
Recommended administered
activity (based on weight only)
in Children and ADOLESCENTSa,b
(Continued)
EANM Paediatric Does Card (2007

version) may be also be used.
c
Higher administered activity of 37 MBq

(1 mCi) may be considered for neonatal
jaundice. cEANM Paediatric Does Card
(2007 version) may be used.
Administered activities at left assume that

image data are reframed at 1 min/image.
Administered activity may be reduced if
image data are reframed at longer time
per image. cEANM Paediatric Dose Card
(2007 version) may also be used.
Low end of dose range should be

considered for smaller patients.
Administered activity may take into
account patient mass and time available
on PET scanner. cEANM Paediatric Does
Card (2007 version) may also be used.
c
version) may also be used.

version) may also be used in patients
weighing more than 10 kg
c
Comments
Appendix F-3BNorth American Consensus Guidelines for Administered Radiopharmaceutical Activities

979
Tc-pertechnetate (Meckel
diverticulum imaging)
Radiopharmaceutical
Tc (for cystography)
No weight-based dose
For solid gastric emptying
9.25 MBq (0.25 mCi)
9.25 MBq (0.25 mCi)
18.5 MBq (0.5 mCi)
9.25 MBq (0.25 mCi)
c
EANM Paediatric Dose Card (2007
version) may also be used.
Comments
18.5 MBq (0.5 mCi)
37 MBq (1.0 mCi)
Administered activity will depend on age

of child, volume to be fed to child, and
time per frame used for imaging.
99m
Tc-sulfur colloid is usually used to label
egg.
99m
No more than 37
Tc-sulfur colloid, 99mTc-pertechnetate,
99m
MBq (1.0 mCi) for
Tc-diethylene triamine pentaacetic
each bladder-filling acid, or possibly other 99mTc
cycle
radiopharmaceuticals may be used.
There are a wide variety of acceptable
administration techniques for 99mTc,
many of which will work well with lower
administered activities.
Maximum
Minimum administered administered
activity
activity
This information is intended as a guideline only. Local practice may vary depending on patient population, choice of collimator, and specific requirements of clinical protocols.
Administered activity may be adjusted when appropriate by order of the nuclear medicine practitioner. For patients who weigh more than 70 kg, it is recommended that maximum
administered activity not exceed product of patients weight (kg) and recommended weight-based administered activity. Some practitioners may choose to set fixed maximum
administered activity equal to 70 times the recommended weight-based administered activity, for example, approximately 370 MBq (10 mCi), for 18F body imaging. The administered
activities assume the use of a low energy high resolution collimator for Tc-99m radiopharmaceuticals and a medium energy collimator for I-123-MIBG. Individual practitioners may use
lower administered activities if their equipment or software permits them to do so. Higher administered activities may be required in certain patients. No recommended dose is given
for 67Ga-citrate. Intravenous 67Ga-citrate should be used infrequently and only in low doses.
b
Reproduced with permission from: M. Gelfand, M. Parisi, T. Treves: Pediatric Radiopharmaceutical Administered Doses: 2010 North American Consensus Guidelines J Nucl Med Vol. 52,
No. 2 February 2011
c
The European Association of Nuclear Medicine (ENAM) Paediatric Dose Card (2007 Version) Reference: Lassmann M, Biassoni L, Monsieurs M, Franzius C, Jacobs F. The new EANM
paediatric dosage card. 2007; 34:796798. Additional notes and erratum related to the Lassmann reference can found in Eur J Nucl Med Mol Imaging. 2008;35:16661668 and
Eur J Nucl Med Mol Imaging. 2008;35:2141.
99m
Tc-sulfur colloid
For oral liquid gastric
emptying
99m
18
F-sodium fluoride
99m
Recommended administered
activity (based on weight only)
in Children and ADOLESCENTSa,b (Continued)
Appendix F-3BNorth American Consensus Guidelines for Administered Radiopharmaceutical Activities
980
0.13
0.017
0.005
0.009
0.003
Tc-99m Disofenin also known

as HIDA (iminodiacetic acid)
Tc-99m DMSA (dimercaptosuccinic acid) also known as

Succimer

known as Ceretec and HMPAO
albumin (MAA)
0.266
0.072
0.082
I-131 Sodium iodide
In-111 Pentetreotide also

known as Octreoscan
0.925
0.02
0.25
I-123 Sodium iodide
I-125 Albumin
0.081
0.01
0.032
0.019
0.0629
0.481
0.303
0.074
0.344
0.022
0.081
0.093
rad/mCi
3.7
F-18 Fluoro-deoxyglucose
0.022
Co-57 Vitamin B12 also known

as Schillings test
Ga-67 Citrate
mGy/MBq
1.0
RADIOPHARMACEUTICAL
Early
0.004
0.007
0.005
0.0150
0.096
0.06
0.068
0.078
0.014
0.2
0.022
0.017
0.68
mGy/MBq
0.015
0.025
0.017
0.056
0.355
0.222
0.252
0.289
0.052
0.74
0.081
0.063
2.516
rad/mCi
3 Months
0.005
0.005
0.004
0.012
0.096
0.035
0.23
0.038
0.011
0.18
0.017
0.008
0.84
mGy/MBq
0.019
0.018
0.015
0.044
0.355
0.13
0.851
0.141
0.041
0.666
0.063
0.028
3.108
rad/mCi
6 Months
DOSE AT DIFFERENT STAGES OF GESTATION
THE MOTHER FOR SOME COMMONLY USED RADIOPHARMACEUTICALSa
0.004
0.004
0.003
0.007
0.094
0.031
0.27
0.026
0.01
0.13
0.017
0.007
0.88
mGy/MBq
(Continued)
0.015
0.013
0.013
0.025
0.348
0.115
0.999
0.096
0.036
0.481
0.063
0.025
3.256
rad/mCi
9 Months
APPENDIX F-4ASUMMARY OF ABSORBED DOSE ESTIMATES TO THE EMBRYO/FETUS PER UNIT ACTIVITY ADMINISTERED TO

981
0.00152
0.359
0.00005
0.058
0.003
0.007
0.002
0.022
0.01
0.012
0.004
0.007
0.009
0.008
0.014
0.005
mGy/MBq
0.00018
0.215
0.01
0.026
0.008
0.081
0.035
0.044
0.016
0.024
0.032
0.03
0.052
0.02
rad/mCi
3 Months
0.00004
0.047
0.003
0.005
0.003
0.014
0.007
0.008
0.003
0.004
0.004
0.004
0.006
0.003
mGy/MBq
0.00013
0.174
0.011
0.02
0.012
0.052
0.026
0.031
0.012
0.013
0.015
0.014
0.02
0.01
rad/mCi
6 Months
0.00003
0.027
0.003
0.004
0.004
0.009
0.004
0.005
0.003
0.003
0.005
0.004
0.005
0.002
mGy/MBq
0.0001
0.1
0.01
0.013
0.014
0.034
0.016
0.02
0.01
0.011
0.017
0.013
0.019
0.009
rad/mCi
9 Months
Source: Stabin M.G., Blackwell R., Brant R.L., Donnelly E., Kinf V.A., Lovins K., Stovall M. Fetal Radiation Dose Calculations. ANSI N13.54 2008, Washington, DC: American National
Standards Institute, 2008.
0.00041
Xe-133 Xenon gas (rebreathing

for 5 min)
0.014
0.004
0.097
Tc-99m WBC
Tl-201 Thallous chloride (rest)
0.036
0.01
Tc-99 Tetrofosmin also known

as Myoview
0.041
0.007
0.011
0.002
0.044
0.056
0.024
0.022
0.044
0.041
0.067
Tc-99m Sodium pertechnetate
0.012
Tc-99 Sestamibi also know as

Cardiolite (stress)
rad/mCi
0.023
0.015
Tc-99 Sestamibi also known as

Cardiolite (rest)
Tc-99m Pentetate also known

as Tc-99m DTPA
0.006
0.012
Tc-99m Bicisate also known as

ECD and Neurolite
0.006
0.011
Tc-99m Mertiatide also known

as MAG3
0.018
Tc-99m Medronate also known

as Tc-99m Methylene diphosphonate (MDP)
mGy/MBq
0.006
RADIOPHARMACEUTICAL
Early
DOSE AT DIFFERENT STAGES OF GESTATION
THE MOTHER FOR SOME COMMONLY USED RADIOPHARMACEUTICALSa (Continued)
APPENDIX F-4ASUMMARY OF ABSORBED DOSE ESTIMATES TO THE EMBRYO/FETUS PER UNIT ACTIVITY ADMINISTERED TO
982
983
Appendix F-4BEffective Dose to the newborn and infant per unit
activity administered FROM THE MOTHERS BREAST

MILKa
Radiopharmaceutical
67
Newborn mSv/MBq (rem/mCi)
1-y-old mSv/MBq (rem/mCi)
1.2 (4.4)
0.490 (1.81)
Tc-DTPA
0.030 (0.111)
0.014 (0.052)
Tc-MAA
0.17 (0.63)
0.068 (0.252)
Tc-pertechnetate
0.14 (0.52)
0.062 (0.229)
Ga-citrate
99m
99m
99m
131
51
I-Nal
5,400 (20,000)
3,900 (14,400)
0.028 (0.104)
0.012 (0.044)
Tc-DISIDA
0.22 (0.81)
0.095 (0.35)
Tc-glucoheptonate
0.080 (0.30)
0.036 (0.13)
Tc-HAM
0.20 (0.74)
0.083 (0.31)
Tc-MIBI
0.14 (0.52)
0.065 (0.24)
Tc-MDP
0.063 (0.23)
0.026 (0.096)
Tc-PYP
0.066 (0.24)
0.028 (0.10)
Tc-RBC in vivo labeling
0.070 (0.26)
0.031 (0.12)
Tc-RBC in vitro labeling
0.071 (0.26)
0.031 (0.12)
Cr-EDTA
99m
99m
99m
99m
99m
99m
99m
99m
Tc-sulfur colloid
0.092 (0.34)
0.042 (0.16)
111
In-white blood cells
5.5 (20)
2.2 (8.1)
123
I-Nal
2.7 (10)
1.9 (7.0)
123
I-OIH
0.051 (0.19)
0.022 (0.081)
123
I-MIBG
2.7 (10)
1.9 (7.0)
125
I-OIH
0.20 (0.74)
0.082 (0.30)
131
I-OIH
0.23 (0.85)
0.093 (0.34)
Tc-DTPA aerosol
0.052 (0.19)
0.022 (0.081)
Tc-MAG3
0.027 (0.10)
0.012 (0.044)
Tc-white blood cells
0.20 (0.74)
0.074 (0.27)
Tl-chloride
3.6 (13)
2.1 (7.8)
99m
99m
99m
99m
201
Effective Dose to infant per unit activity administered intravenously to infant. (i.e., assumes 100% of the
activity ingested by the infant is instantaneously absorbed. Calculation based on ICRP 60 methodology. See
below for narrative and sample calculation of effective dose to the infant based on serial measurements of
activity in the breast milk.
Dose to infants thyroid per unit activity administered intravenously (or orally) to infant.
DTPA, diethylenetriamine pentaacetic acid; MAA, macroaggregated albumin; EDTA, ethylenediaminetetraacetic acid; DISIDA, disofenin (iminodiacetic acid derivative); HAM, human albumin microspheres; MIBI,
methoxyisobutyl isonitrile; MDP, methylene diphosphonate; PYP, pyrophosphate; RBC, red blood cells; WBC,
white blood cells; OIH, orthoiodohippurate; MIGB, metaiodobenzylguanidine; MAG3, mercaptoacetyltriglycine.
Source: Stabin MG, Breitz HB. Breast milk of radiopharmaceuticals: mechanisms, findings, and radiation
dosimetry. J Nucl Med 2000; 41 (5): 863873.
a
984
Narrative and example1
Recommended duration of interruption of breast-feeding following radiopharmaceutical administration to a patient who is nursing an infant or child was provided
in Chapter 21, Table 21-9. The interruption schedules for the nursing infant were
derived using a dose criteria of 1 mSv effective dose to the infant. However, Stabin
and Brietz recommend taking breast milk samples from subjects when possible to
determine, on an individual basis, the best recommendation for the duration of interruption. According to their recommendations in the article cited above, breast milk
samples should be obtained: (1) at about 3 h after administration (this is when
the peak concentrations have most often been observed); (2) then, as many more
samples as the patient is willing and able to give, over 2 to 3 effective half-times of the
radiopharmaceutical in the body. If there is uncertainty about the biologic half-time,
the radionuclide physical half-life may be used to estimate this overall time period.
A minimum of two more samples (after the first sample at 3 h) should be obtained
to calculate a good estimate of the retention half-time in the milk. Once the peak
concentration and rate of decrease of the activity are determined, some approximate
calculations can be performed by any physician or physicist to estimate the amount
of activity that the infant will ingest starting at different points in time. One can set
up a calculation in a simple spreadsheet that sums, for whatever sampling schedule
the mother suggests that the infant is likely to follow, the amounts of activity likely
to be ingested, using the observed concentrations and rate of elimination. Then, the
dose conversion factors in Appendix F-4B can be used to calculate the infant dose.
This technique is illustrated in the following example from the Stabin and Breitz
article: assume that for an administration of Tc-99m pertechnetate the breast-milk
concentration reported at 3 h after administration to the mother is 2 10-2 MBq/mL.
Three more samples, taken over the next 8 h, show a clearance biologic half-time of
20 h. The effective half-time is
(6 h 20 h)/(6 h + 20 h) = 4.6 h
The mother wants to feed the baby (a newborn) approximately every 4 h. The volume
ingested per feeding is assumed to be in feedings of 142 mL every 4 h (i.e., ~850 mL/d).
Thus for the following times, starting at 12 h after administration (we are already at 11
h after administration), the babys intakes for the next seven feedings would be
T (h)
At(MBq)
12
0.735
16
0.403
20
0.221
24
0.121
28
0.067
32
0.036
36
0.02
40
0.011
Total activity = 1.62 MBq
1
Source: Stalin MG, Breitz HB. Breast milk excretion of radiopharmaceuticals: mechanisms, findings and
radiation dosimetry. J Nucl Med 2000;41(5):863873.
985
Each value of At is given by the expression:

At = (142 mL 0.02 MBq/mL) e0.693 (T 3)/(4.6)
At (MBq) is the activity ingested by the infant at the feeding at time T (h). We are assuming
that the peak concentration (0.02 MBq/mL) occurred at 3 h and then decreased with
the effective half-time (4.6 h) thereafter. We took the calculations out to 40 h, when the
concentration seemed to have diminished to the point that further contributions would
be negligible. The sum of the activity values listed previously is 1.62 MBq. In Appendix
F-4B, we find a dose value for Tc-99m pertechnetate of 0.14 mSv/MBq for a newborn.
The cumulative dose, assuming that feeding started at 12 h, would be 1.62 MBq
0. 14 mSv/MBq = 0.23 mSv.
This dose is within the 1 mSv guideline used here, and one would conclude
that breast-feeding could resume safely at 12 h after administration. If the dose had
turned out to be too high, the calculation could be repeated easily, simply excluding
some of the values in the table shown above from the sum, starting at 16 h, then
at 20 h, and so on, until an acceptable dose value was obtained. The time at which
this value was obtained would represent the time at which breast-feeding could be
resumed.
986
Appendix F-4CBreast dose from radiopharmaceuticals excreted
in breast milk
Radiopharmaceutical
Administered activity in MBq (mCi)
Breast dose (Gy)

BEST case
WORST case
Ga-citrate
185 (5.0)
2.18E-04
1.10E-02
Tc-DTPA
740 (20)
6.09E-06
1.20E-04
Tc-MAA
148 (4)
1.55E-05
1.21E-03
Tc-pertechnetate
1110 (30)
1.86E-05
2.52E-03
67
99m
99m
99m
I-Nal
5550 (150)
1.96E+00
1.85 (0.05)
4.21E-09
2.52E-08
Tc-DISIDA
300 (8)
1.94E-05
5.98E-05
Tc-glucoheptonate
740 (20)
3.58E-05
7.40E-05
Tc-HAM
300 (8)
8.48E-06
2.33E-04
Tc-MIBI
1110 (30)
5.54E-06
5.09E-05
Tc-MDP
740 (20)
2.69E-05
3.76E-05
Tc-PYP
740 (20)
4.16E-05
2.26E-04
Tc-RBC in vivo
740 (20)
2.46E-06
1.14E-03
Tc-RBC in vitro
740 (20)
9.25E-06
1.61E-05
131
Cr-EDTA
51
99m
99m
99m
99m
99m
99m
99m
99m
Tc-sulfur colloid
444 (12)
3.17E-05
4.64E-04
In-WBCs
18.5 (0.5)
5.03E-06
2.52E-05
I-Nal
14.8 (0.4)
I-OIH
74 (2)
7.50E-05
5.84E-04
I-MIBG
370 (10)
2.71E-04
I-OIH
0.37 (0.01)
8.46E-07
I-OIH
11.1 (0.3)
4.97E-05
3.22E-04
Tc-DTPA aerosol
37 (1)
1.22E-07
2.49E-06
Tc-MAG3
185 (5)
3.04E-06
6.01E-05
Tc-WBCs
370 (10)
1.11E-04
1.51E-02
Tl-chloride
111 (3)
2.35E-05
4.14E-05
99m
111
123
123
123
125
131
99m
99m
99m
201
4.74E-04
Best and worst case as observed from the literature.

DTPA, diethylenetriamine pentaacetic acid; MAA, macroaggregated albumin; EDTA, ethylenediaminetetraacetic
acid; DISIDA, disofenin (iminodiacetic acid derivative); HAM, human albumin microspheres; MIBI, methoxyisobutyl
isonitrile; MDP, methylene diphosphonate; PYP, pyrophosphate; RBC, red blood cells; WBC, white blood cells;
OIH, orthoiodohippurate; MIGB, metaiodobenzylguanidine; MAG3, mercaptoacetyltriglycine.
From Stabin MG, Breitz HB. Breast milk excretion of radiopharmaceuticals: mechanisms, findings, and radiation
dosimetry. J Nucl Med. 2000;41(5):863873.
a
APPENDIX
Convolution and Fourier

Transforms
G.1 Convolution
Convolution is the mathematical property that describes the blurring process in
medical imaging, among other physical phenomenon. It is the basis of optical physics
and imaging physics. For a function f(x) and a convolution kernel h(x), convolution
is performed as below resulting in the function g(x):
g(x ) =
f ( x ') h( x x ') dx '
[G-1a]
where x is a dummy variable used for integration. Convolution is also the basis
of many image processing procedures. The shorthand symbol for convolution is
described in Equation G-1b:
g ( x ) = f ( x ) h( x )
[G-1b]
where is the mathematical symbol for convolution. The convolution process can
be extended to two dimensions:

g ( x, y ) =
y ' =
x ' =
f ( x ', y ') h( x x ', y y ') dx ' dy '
[G-1c]
Figure G-1A shows a one-dimensional convolution, and it is straightforward to extend

convolution to a three-dimensional function as well. With the advent of near-isotropic
three-dimensional image data sets, three-dimensional convolution techniques are used
with increasing utility for many imaging applications.
Figure G-1 illustrates the effect of a five element rectangular (RECT) convolution kernel, h(x), on an input signal f(x) with some considerable noise. The output function g(x)
is much smoother than the input signal, but the edges are not as sharp. The RECT kernel
is illustrated in the inset. A RECT kernel used in convolution is also known as a boxcar
average. If the kernel is normalized to unit area as in this case (5 0.2 = 1), the total
amplitude of the convolved function will be equal to that of the input function. Convolution kernels that have all positive values (as in Fig. G-1) will always result in some degree
of smoothing of the input function. Let [0.2, 0.2, 0.2, 0.2, 0.2] represent the RECT kernel
(with highlighted color being the center element). Other kernels with all positive elements such as [0.1, 0.25, 0.4, 0.25, 0.1] will also result in smoothing, for example.
Convolution can be used to enhance edges as well, and Figure G-2 illustrates this
with a kernel designated as [0.0, 0.5, 2.0, 0.5, 0.0] (see inset). In this case, the
noise in the input function f(x) is made worse by the convolution procedure resulting
in the output function g(x). Although the noise in the signal is amplified, the edges
are enhanced as well. Edge enhancement (also known as high pass filtering) is a key
part of filtered backprojection reconstruction in tomographic imaging applications.
The convolution operation can perform a number of interesting mathematical
procedures on an input signal. Figure G-3 shows a convolution procedure with a
987
988
Amplitude (A.U.)
120
100
Kernel value
140
80
h(x)
0.25
0.20
0.15
0.10
0.50
0.00
0.50 0.30 0.10 0.10 0.30 0.50
x axis
g(x)
60
f(x)
40
20
0
10
2
0
2
Position (mm)
10
FIGURE G-1 This figure illustrates the impact of convolution with a noise-suppressing kernel. The original
signal f(x) was convolved with the five element RECT kernel h(x) shown in the inset, resulting in the smoothed
function g(x).
kernel [0.0, 0.0, 1.0, 1.0, 0.0]. In this case, the output function g(x) is the discrete
derivative of the input function f(x). Notice that in this example the sum of the kernel
is zero, and the average value of the output function g(x) also becomes zero.
The convolution operation has a number of shifting and scaling properties.
Figure G-4 illustrates a convolution kernel with eleven elements, all zeros with the
tenth element set to 0.5: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0.5, 0]. By placing the only nonzero element away from the center of the kernel, the output function g(x) is shifted
laterally relative to the input function. By having only a negative element, the entire
output function g(x) has its polarity flipped relative to the input function f(x). Finally,
because the amplitude of the kernel has a magnitude of 0.5, the output function is
scaled to be 50% of the amplitude of the input function (albeit in the negative direction due to the negative value of the kernel).
160
120
2.50
2.00
1.50
1.00
0.50
0.00
0.05
0.50 0.30 0.10 0.10 0.30 0.50
x axis
Kernel value
Amplitude (A.U.)
140
100
h(x)
edges
80
f(x)
60
40
20
0
10
g(x)
8
10
Position (mm)
FIGURE G-2 The role of an edge-enhancing kernel, h(x), is illustrated. The original signal f(x) was convolved
with h(x), resulting in the noisier function g(x). While g(x) is noisier than the original function, the two edges
(arrows) have also been enhanced.
Appendix G Convolution and Fourier Transforms
989
120
Amplitude (A.U.)
100
80
Kernel value
140
60
1.50
1.00
0.50
0.00
0.50
1.00
1.50
0.50 0.30 0.10 0.10 0.30 0.50
x axis
40
20
0
20
40
10
10
Position (mm)
FIGURE G-3 This figure illustrates the original signal (blue) with only a small amount of noise, and the
kernel (inset) was used to compute the derivative signal (red). The numerical derivative was computed using
convolution.
G.2 The Fourier Transform

In the analysis of time-based signals, the Fourier transform converts from the time
domain (x-axis labeled in seconds) to the temporal frequency domain (x-axis labeled
in sec1). While time-series analysis has applications in radiological imaging in ultrasound Doppler and magnetic resonance imaging, here we focus on input signals that
are in the spatial domain (x-axis labeled typically in mm), which is pertinent to any
anatomical medical image. The Fourier transform of a spatial domain signal converts
it to the spatial frequency domain (x-axis labeled in mm1).
The one-dimensional Fourier transform of a spatial domain signal input function
f(x) is given by
F ( ) =
f ( x ) e 2 i x dx
[G-2a]
120
100
80
60
60
20
Kernel value
Amplitude (A.U.)
140
0.00
0.20
0.40
0.60
0.80
1.00
0.200.00 0.20 0.40 0.60 0.80 1.00
x axis
flip
0
20
40
60
80
10
scaling
lateral shift
6
2
0
2
Position (mm)
10
FIGURE G-4 This figure illustrates three different effects of convolution(1) because the total amplitude
of the kernel was negative, the convolved function is flipped downward relative to the input signal (i.e., it has
become negative); (2) because the amplitude of the kernel was 0.5, the convolved function has half of the
(negative) amplitude as the original; and (3) because the d-function used (see kernelinset) was shifted from
zero, the convolved function is shifted laterally compared to the original function.
990
where F(v) is the Fourier transform of f(x) and is in the units of amplitude versus
spatial frequency (mm1), and where i is equal to 1 . As with convolution, the
Fourier transform can be performed in two dimensions as described by

F (, ) =
y =
x =
f ( x , y ) e 2 i ( x + y ) dx dy
[G-2b]
Three-dimensional Fourier transforms can also be performed. For simplicity, only

one-dimensional Fourier transforms are described below. The shorthand notation for
the Fourier transform in Equation G-2a is given by

F ( ) = FT [ f ( x )]
[G-3a]
Of course, other functions such as h(x) can be transformed as well:

H ( ) = FT [h( x )]
[G-3b]
Multiplication of the two frequency domain functions F() and H() is the equivalent
of convolution in the spatial domain,

G( ) = F( ) H( )
[G-4]
except that the product G() then needs to be converted back to the spatial domain
using the inverse Fourier transform (shorthand FT1[]):

g(x ) =
G ( ) e 2 ix d
[G-5]
Notice that the use of the Fourier transform and its inverse in Equations G-3a, G-3b,
G-4, and G-5 provides the same result as the convolution function described in
Equation G-1a. Explicitly:

g ( x ) = f ( x ) h( x ) = FT 1 {FT [ f ( x )] FT [h( x )]
[G-6]
For most applications, the implementation of these equations in a computer is faster

using Fourier transforms* than using convolution, and this has led to the widespread
use of the Fourier transform in image processing, image reconstruction, and other
image analysis operations.
G.3 The Fourier Transform in Filtered Backprojection

The purpose of this section is to tie the discussion of convolution and Fourier transforms into previous discussions on filtered backprojection in tomographic reconstruction (Chapters 10 and 19). The backprojection procedure results in a characteristic
1/r blurring phenomenon, and the filtering procedure in filtered backprojection is
the mathematical operation described in Equation G-6 that corrects for this 1/r blurring. In the case of filtered backprojection, h(x) is a convolution kernel that corrects
for the 1/r blurringand hence it can be called a deconvolution kernel. Deconvolution simply refers to convolution when it is used to correct for a specific effect (such
as blurring). In the image reconstruction literature, it is common to talk about the filtering function h(x) in the frequency domain, that is, FT(h(x)) or H(). As was seen in
Chapter 10, h(x) is an edge-enhancing kernel, more complicated (and wider in terms
*The Fast Fourier Transform is the principal tool used for computer implementation of the Fourier Transform, as it achieves computational efficiency over a general Discrete Fourier Transform.
991
of number of elements) but similar conceptually to that shown here in Figure G-2.
In the frequency domain, H(n) is a ramp filter that in clinical CT imaging always has
an additional high-frequency roll-off term as well. Thus, H(n) = R(n) A(n), where R()
is the RAMP filter (thus R(n) = an, where a is a scalar) and A(n) is an apodization
filter that approaches zero as n approaches the Nyquist frequency.
G.4 The Fourier Transform and the Modulation Transfer
Function
The rigorous assessment of spatial resolution as discussed in Chapter 4 involves

the measurement of the modulation transfer function (MTF) of an imaging system.
There are several approaches to this, but the most widely used method is to measure
the line spread function 1(x), and from it, the MTF(f) is computed. In this section,
more details about this process are provided. The line spread function (LSF) 1(x)
is measured or synthesized as discussed in Chapter 4. Here we show the analytic
description of MTF calculation, but in most practical situations, computer programs
are used that perform the mathematics in a discrete mannerusing a computer.
The first step of MTF assessment is to normalize the area of 1(x) to unity:
1=
l ( x ) dx
[G-7]
The Fourier transform of the line spread function is then computed, resulting in the
optical spread function, OTF(f).

OTF ( f ) =
l ( x ) e 2 ifx dx
[G-8]
The line spread function is a real function, whereas the OTF(f) is a complex function;
that is, it has real and imaginary components. As a reminder, a complex number c has
a real component a and an imaginary component b, where c = a + bi, and as mentioned earlier, i = 1 . The MTF(f) is computed as the modulus of the OTF(f):

{OTF ( f )}2 + {OTF ( f )}2
MTF ( f ) =
[G-9]
where {OTF ( f )} is the real component of the OTF and {OTF ( f )} is the imaginary component. With this transformation, the MTF(f) represents the modulus of the
OTF and as such is a real function. Another way to write Equation F-9 is

MTF ( f ) = OTF ( f )
[G-10]
Because of the properties of the Fourier transform, since 1(x) is normalized to 1.0,
the MTF(f) value at f = 0 is also unity, that is, MTF(0) = 1.
Fourier Transform Pairs

A set of three line spread functions is shown in Figure G-5. These 1(x) plots corresponding loosely to those for three imaging modalitiesanalog screen film radiography (curve
A), computed tomography (curve B), and single photon emission computed tomography (SPECT, curve C). The full width at half maximum of these curves is approximately
0.14, 0.6, and 6 mm, respectively. The MTFs of these three 1(x) curves are shown in
Figure G-6. It can be seen that the widest 1(x) (curve C) in Figure G-5 has the worst MTF
in Figure G-6. Note that the limiting resolution for the three MTFs shown in Figure G-6
(at the 5% dashed line) corresponds approximately to 0.2, 1.4, and 6.6 mm1.
Amplitude (A.U.)
992
110
100
90
80
70
60
50
40
30
20
10
0
B
A
Position (mm)
FIGURE G-5 A number of gaussian functions are illustrated (A, B, and C). This figure shows the gaussian
functions in the spatial domain, and these functions could represent line spread functions.
Gaussian
The Fourier transform of a gaussian is also a gaussian. Because the line spread function in many settings is well approximated by a gaussian function, that example will
be used here to illustrate the utility of Fourier transform pairs.
Let the spatial domain function 1(x) be given by
l ( x ) = e ( x
)/ a 2
[G-11a]
where a is a constant that affects the width of the line spread function. The analytical
Fourier transform of Equation G-11a is given by
2 2
MTF ( ) = a e a
[G-11b]
where the value of a in Equation G-11b is the same as that in Equation G-11a for
the Fourier transform pair. The LSF/MTF curves shown in Figures G-5 and G-6
(respectively) were produced with values of a corresponding to 0.15, 0.68, and 6.0
(for curves A, B, and C as labeled in these figures).
MTF(f)
110
100
90
80
70
60
50
40
C
30
20
10
0
0
10

FIGURE G-6 These gaussian functions are the Fourier transforms of the gaussian functions seen in Figure
G-5. Notice that the widest gaussian in Figure G-5 results in the narrowest function on this plot. While the
relationship between the gaussian functions shown in Figure G-5 and their FT counterparts in this figure is
general and illustrates Fourier transform pairs, if the functions in Figure G-5 were line spread functions, then
the functions illustrated here in Figure G-6 are the corresponding MTFs.
993
The LSF/MTFs shown in Figures G-5 and G-6 were meant to illustrate the use
of the Fourier transform in image science, but they also illustrate gaussian Fourier
transform pairs. It should be noted that gaussian FT pairs are not limited to LSFs and
MTFs, but rather this is a general relationship. In addition, there are many LSFs and
MTFs that do not fit a gaussian function.
RECT
x
The rectangle function, or RECT , is a commonly used function in image scia
x
encein the absence of other sources of blur, the RECT function describes how
a
x
a detector element (dexel) of width a integrates a signal incident upon it. RECT
a
is defined by
a
x
RECT = 1, x <
a
2
= 0, elsewhere
[G-12]
With this definition, the limits of integration of the Fourier transform can be modified resulting in the following analytical derivation:
F ( ) =
x
RECT e 2 i x dx =
a
a /2
a / 2
e 2 i x dx =
sin( a )

[G-13]
The SINC function is defined as

SINC ( x ) =
sin( x )

x
[G-14]
And thus Equation G-13 can be written as

F ( ) = a SINC (a )
[G-15]
x
The above derivation illustrates that RECT and SINC(n) are Fourier transform pairs.
a
The SINC function is shown in Figure G-7, for different RECT apertures. As the aperture
(a) gets smaller, the RECT function (detector element in this example) gets narrower, and
MTF (f)
110
MTF(f)
100
90
80
70
60
50
110
100
90
80
70
60
50
40
30
20
10
0
40
a = 0.20 mm
30
20
1 / [2a]
aliasing
1/a
4
5
1
2
3
a = 0.10 mm
a = 0.05 mm
10
0
0
Spatial Frequency
10
(mm1)
FIGURE G-7 The analytical Fourier transform of a RECT function is SINC, and three SINC functions are illustrated in this figure. The SINC is the theoretically best physical MTF that can be achieved with a rectangularshaped detector element with a given width a.
994
the spatial resolution (MTF) improves. Figure G-7 illustrates the MTF curves for rectangular detector element apertures corresponding to 0.05, 0.10, and 0.20 mm in width
(value of a). Indirect detector systems that use an x-ray phosphor have other significant
sources of resolution loss (optical blur), but for direct x-ray detectors, the RECT function
characterizes the detectors response to the incident x-ray beam and thus the SINC function approximates the MTF of indirect detector systems. An MTF described by SINC
reaches MTF(f) = 0 at 1/a, whereas the Nyquist frequency (see Chapter 4) is given by Fn =
1/(2a). This is shown in the inset in Figure G-7. For direct detector x-ray systems, the near
SINC MTF results in potential signal intensity up to 1/a, whereas all frequencies above
1/(2a) will be aliased.
Digital Sampling and Aliasing

A linear detector array with center-to-center spacing between dexels of D can be
thought of as an array of delta functions (-functions) separated by a distance D
(Fig. G-8). Mathematically, a comb of -functions describes this:
s( x ) =
n
j =0
( x j )
[G-16]
It is noted that (0) = 1, and (x) = 0 when x 0. Using (for example) a rectangular
x
sampling aperture of width a, RECT , the digital sampling function, g(x), is the
a
convolution between these terms:
x
g ( x ) = s ( x ) RECT
a
[G-16]
Figure G-8A illustrates the comb of -functions with sampling interval D. Figure G-8B
shows two different RECT functions, of width a = D and b = 2D. Figure G-8C
illustrates the combined effects of the aperture a and sampling interval D, as described
A.Sampling, s(x)
j=0
B.signal integration
j=1
j=n
RECT
RECT
C.g(x), when a =
D.g(x), when b = 2
FIGURE G-8 A Illustrates the sampling of a detector system with a center-to-center spacing of D, represented as a comb of d-functions. B shows the RECT aperture detector elements of width a (where a = D, and
widith b, where b = 2 D). A contiguous detector array is shown in (C) where D = a, and there are real examples
of this in direct detection x-ray imaging. The function shown in (D) shows a nonphysical detector where the
aperture width b = 2D. The consequences of this situation are shown in Figure G-9B.
995
by Equation G-16. Figure G-8D illustrates the shape of g(x) when the sampling comb
x
is convolved with the aperture function given by RECT .
b
Figure G-9A shows the frequency domain interpretation of Figure G-8C, the Fourier transform of g(x) in Equation G-16. The comb of d-functions in the spatial domain
separated by a sampling interval of D becomes a comb of d-functions separated by 1/D
in the frequency domain. As seen in Figure G-7, the SINC function has nonzero amplitude to 1/a, and so when a = D as in Figure G-9A, the MTF values overlap between
samples and this means that aliasing can occur. Although the sampling scenario illustrated in Figure G-8D is not practical physically (except as proposed by Rowlands),
it represents an optimal approach to the management of aliasing. Because the wider
detector element results in more blurring of the detected signal, the magnitude of the
MTF is degraded at higher spatial frequencies such that it reaches zero at the Nyquist
frequency, and this is shown in Figure G-9B. In this situation, aliasing cannot occur
because signal amplitude becomes zero above the Nyquist frequency. While aliasing is
eliminated, the MTF is reduced and therefore the spatial resolution of the imaging system is degraded relative to that shown in Figure G-9A. Aliasing is most problematic in
terms of a radiological signal when that signal is periodic in nature, such as the x-ray
shadow of an antiscatter grid. It is recognized that Figure G-9A also allows periodic
components of image noise to be aliased, meaning that high frequency image noise
will wrap around the Nyquist frequency and increase noise levels in the image.
It is noted that in the case of indirect x-ray imaging systems that use an x-ray scintillator, the optical blurring that occurs in the system degrades the MTF at high spatial frequencies such that little if any signal amplitude exists above the Nyquist frequency. The
influence of the optical blurring is to increase the effective aperture width, similar to
that illustrated in Figure G-8D, although the blurring function is not RECT in shape.
G.5 Fourier Analysis of Noise: The Noise Power Spectrum

Image noise is measured in the absence of an object, so typically a constant image is
produced by the imaging modality. Examples in radiography, mammography, or fluoroscopy would be air scans (no object), and in CT, a homogeneous cylinder of plastic
1/
when a =
1/a
when b = 2
1/
1/b
FIGURE G-9 A. This figure shows

the frequency domain consequences of
the detector defined in Figure G-8Ca
RECT aperture a sampled with a pitch
of D where a = D B. This shows the frequency domain construct of function g(x)
shown in Figure G-8D, where b = 2D.
With this situation, the MTF approaches
zero amplitude at the Nyquist frequency,
eliminating the potential for aliasing.
996
can be scanned. Every effort is made to make these images as flat as possibleone
approach to flatten the image is to subtract two independently acquired images,
and this reduces spatial bias due to the heel effect, inverse square law, or pincushion
distortion effects. The variance of the noise on the subtracted image is divided by
a factor of two in order to compensate for the noise propagation when two images
convolution with gaussian
white noise
fr
fx
correlated noise
fy
spatial domain noise
2D noise power spectra
2D Fourier Transform
A
10000
9000
correlated noise
8000
NPS (f)
7000
6000
5000
4000
white noise
3000
2000
1000
0
0
FIGURE G-10 A.The spatial domain (left) and two-dimensional NPS(fx, fy) functions for two examples of
noise are shown, white noise on the top row and correlated noise on the bottom row. Most imaging systems
demonstrate correlated noise. The correlation was produced in this example by convolving the white noise
example with a two-dimensional gaussian function. B. The two-dimensional NPS shown on the right side of
(A) is radially symmetric, and this allows it to be integrated radially. Radial integration converts the function
to a one-dimensional function, which is a function of frequency, as shown in (B). Here, two one-dimensional
NPS(f) functions are shown for the two-dimensional NPS(fx, fy) shown in (A) (top and bottom right side). The
NPS(f) functions shown in this figure characterize the noise texture on an image.
997
are subtracted. In most settings, however, the NPS is measured on unsubtracted but
sometimes detrended images. The NPS is computed as

NPS ( f )
(I ( x ) I )
e 2 ifx dx
[G-17]
The integral of the NPS over all frequencies is the noise variance:

2 = NPS ( f ) df
[G-17]
The variance describes the overall magnitude of the noise in an image, while the
NPS(f) is a measure of the noise texture, that is, the frequency-dependent character of the noise. Figure G-10 illustrates differences in noise texture. Figure G-10A
illustrates white noise in the spatial domain, and the accompanying NPS(f) is white
as well. Figure G-10B illustrates noise that has a correlation structure in it, both in
the spatial and the frequency domains. The spatial noise pattern in Figure G-10B
was produced by convolving the noise pattern in Figure G-10A with an 11 11
gaussian. The NPS(f) in Figure G-10B shows a bell shape, and indeed when radially
averaged (Fig. G-10C), the NPS(f) appears gaussian-like.
Suggested Reading
Barrett HH, Sindell W. Radiological imaging: the theory of image formation, detection, and processing, vol. 1,
New York, NY: Academic Press, 1981.
Dainty JC, Shaw R. Image science. London: Academic Press, 1974.
Ji WG, Zhao W, Rowlands JA. Digital x-ray imaging using amorphous selenium: reduction of aliasing. Med
Phys 1998;25(11):21482162.
APPENDIX
Radiation Dose: Perspectives

and Comparisons
Table H-1 provides the ICRP tissues and tissue weighting factors (wT) that were used
in the calculation of effective dose equivalent (HE) in ICRP report 26 (ICRP 1977)
and effective dose (E) in ICRP report 60 (ICRP 1990) and ICRP report 103 (ICRP
2007). Note that many of the effective dose estimates provided in Appendix E refer
to references where the effective dose cited was that calculated using the previous
(i.e., ICRP 60) system for effective dose and their associated tissue weighting factors.
In some cases, the application of the ICRP 103 system to examinations provided in
those tables will result in different (higher or lower) effective dose estimates.
Figure H-1 shows the relationships between the emission of radiation and the various quantities described in previous chapters to express radiation exposure; kerma;
absorbed dose; equivalent dose (from which the dose to tissues and organs can be
used along with other risk-modifying variables such as age at time of exposure and
gender to estimate cancer incidence and mortality risk to an individual); and effective
dose to the body, which can be used to estimate the stochastic health risk for a group
that has the same general age and gender distribution as the general population.
Figure H-2 provides an example of estimated effective doses and weighted organ
equivalent doses from some standard cardiac radionuclide and CT diagnostic studies
using ICRP report 103 (2007) and ICRP report 60 tissue weighting factors.
Figure H-3 provides a comparison of estimated effective doses (mSv) for standard
myocardial perfusion imaging protocols, determined with the use of ICRP and manufacturers package insert dose coefficients.
Tables H-2 to H-4 and Figures H-4 and H-5 are presented to help put radiation
exposure into perspective by providing examples of other sources of radiation, the
doses from those sources as well as other public health risks, and their relative magnitude. Table H-2 provides information on doses from several sources of public and
occupational radiation exposure. Table H-3 presents the estimated loss of life expectancy from several health risks including an estimate for the risk from background
radiation. It is important to be aware of the assumptions incorporated into the values presented and limitations inherent in making risk comparisons (see table footnotes). Table H-4 provides some examples of the amount of radioactivity in various
natural and artificially produced sources. Figure H-4 illustrates the variation in the
annual effective dose from cosmic radiation in North America. Figure H-5 compares
the average annual effective dose from natural background radiation sources in the
United States to that of a number of European countries.
998
Appendix H Radiation Dose: Perspectives and Comparisons
999
TABLE H-1ICRP TISSUE WEIGHTING FACTORS wT (ICRP 1977; ICRP 1990;
ICRP 2007)
ICRP REPORT
ICRP 26
ICRP 60
ICRP 103
ICRP 26
YEAR
1977
1990
2007
Tissues Included in Remainder Tissues
Quantity
EDE
ED
ED
Adipose tissue
Tissue
Tissue Weighting Factors, wT
Adrenals
Adrenals
Gonads
0.25
0.2
0.08
Brain
STWF
Breast
0.15
0.05
0.12
Connective tissue
Red bone marrow
0.12
0.12
0.12
Extrathoracic airways
Lung
0.12
0.12
0.12
Gall bladder
Thyroid
0.03
0.05
0.04
Brain
IMT
Bone surfaces
0.03
0.01
0.01
Heart wall
Colon
0.12
0.12
Kidney
Kidney
Stomach
0.12
0.12
Lymphatic nodes
Bladder
0.05
0.04
Muscle
Muscle
Esophagus
0.05
0.04
Liver
STWF
STWF
Liver
0.05
0.04
LLI
STWF
STWF
Brain
0.01
Pancreas
0.01
Prostate
Salivary Glands
Skin
ICRP 60
ICRP 103
0.01
0.01
SG
IMT
0.3
0.05
0.12
SI
SI
SI wall
Total
1.0
1.0
1.0
Stomach
STWF
STWF
Spleen
Spleen
Remainder tissues
ULI
Thymus
ULI
STWF
Tissues selected to represent the remainder in ICRP reports 26, 60 and 103 are shown on the right side of the table.
EDE, Effective Dose Equivalent (HE); ED, Effective Dose (E); STWF, See Tissue Weighting Factor Table; LLI, lower
large intestine; SG, salivary glands; SI, small intestine; ULI, upper large intestine.
ICRP 26: ICRP Publication 26, Recommendations of the International Commission on Radiological Protection,
Pergamon Press, Oxford, England (1977).
ICRP 60: International Commission on Radiological Protection. 1990 Recommendations of the International
Commission on Radiological Protection, ICRP Publication 60, Ann.ICRP 21(1-3) (Elsevier, New York), (1991).
ICRP 103: The 2007 Recommendations of the International Commission on Radiological Protection. ICRP Publication 103. Ann ICRP 37, 1332 (2007).
1000
From Exposure to Risk
Radiation source
Outside or Inside
the Body
(e.g., x-rays,
gamma rays,
radioactive
material)
Emission of
Radiation
Exposure Rate (R/hr)

KERMAair Rate
(mGy/sec)
Exposure Rate
Constant
(mGy-m2 GBq1 hr1)
Organ Excess Lifetime Risk

BEIR VII #/105/Sv
based on
Modifying Factors
age at exposure & gender
Estimated
Individual
Lifetime
Risk
#/10,000
Radiation
Tissue
Absorbed weighting Equivalent weighting
Estimated
Doses
Doses
Effective
factors
Assumes
factors
Population
DT (Gy)
HT (Sv)
Dose
wR
general
wT
Excess
to
to
E (Sv)
and
population
Risk
Tissues
Tissues
to the
summation
distribution of #/105/Sv
and
and
Body
age
&
gender
Organs
Organs
Biological
Effectiveness
of the
Radiation
Biological
Sensitivity of the
Tissue to
Stochastic
Effects
(Detrimentprimarily cancer)
Figure H-1 Relationships between the emission of radiation and the various radiation dose quantities
described in Chapter 3 and discussed throughout the text. (Adapted from ICRP. The 2007 Recommendations
of the International Commission on Radiological Protection, ICRP Publication 103. Ann ICRP 2007;37:1332.)
30
Doses determined using ICRP 2007
tissue weighting factors
Effective Dose (mSv)
25
Other
Thyroid
Stomach
Marrow
Lung
Liver
Kidney
Gonads
Esophagus
Colon
Breast
Bladder
20
15
10
Tc-99m
Tc-99m
Tc-99m Thallium Dual lsotope Rb-82
Sestamibi Sestamibi Tetrofosmin
Rest-Stress Stress Rest-Stress
N-13 O-15 water CTCA CTCA with Calcium

without ECTCM Scoring
Ammonia
ECTCM
Figure H-2 Estimated effective doses and weighted organ equivalent doses from cardiac radionuclide and
CT studies. A. Doses determined using ICRP Publication 103 (2007) tissue weighting factors.
30
Doses determined using ICRP 1990
tissue weighting factors
Effective Dose (mSv)
25
Other
Thyroid
Stomach
Marrow
Lung
Liver
Kidney
Gonads
Esophagus
Colon
Breast
Bladder
20
15
10
Tc-99m Tc-99m
Tc-99m Thallium Dual lsotope Rb-82
Sestamibl Sestamibi Tetrofosmin
Rest-Stress Stress Rest-Stress
N-13 O-15 water CTCA CTCA with Calcium

without ECTCM Scoring
Ammonia
ECTCM
Figure H-2 (Continued) B. Doses determined using ICRP Publication 60 (1990) tissue weighting factors.
CTCA indicates 64-slice computed tomography coronary angiogram; CaSc, calcium scoring; and ECTCM,
ECG-controlled tube current modulation. Calculations were performed with ImpactDose (VAMP GmbH, Erlangen, Germany); for Siemens Sensation 64 scanner with retrospective gating, voltage was 120 kVp, pitch 0.2,
and scan length 15 cm. For CTCA, slice thickness was 0.6 mm and tube current-time product was 165 mAs;
ECTCM was simulated by reducing tube current by a third, to 110 mAs. For CaSc, collimation was 20 1.2
mm, and tube current-time product was 27 mAs. Doses shown are arithmetic means of doses to standardized
male and female phantoms. (FromEinstein AJ, Moser KW, Thompson RC, et al. Radiation dose to patients
from cardiac diagnostic imaging. C
irculation 2007;116:12901305.)
Tc-99m Sestamibi
16
16
14
14
12
12
10
10
ICRP
Pl
Tl-201
Tc-99m Tetrofosmin
ICRP
Rb-82
16
30
Pl
14
25
12
20
10
Other
ULI
Thyroid
Stomach
Marrow
Lung
Liver
Kidney
Gonads
Colon
Bladder
15
10
4
5
0
2
ICRP
PI 1
PI 2
PI 3
ICRP
PI
Figure H-3 Comparison of estimated effective doses (mSv) for standard myocardial perfusion imaging
protocols, determined with the use of ICRP and manufacturers package insert dose coefficients. Weighted
equivalent doses were determined with the use of ICRP Publication 60 tissue weighting factors. ULI indicates
upper large intestine. Some notable differences exist between effective doses estimated with the use of ICRP
dose coefficients and those estimated with the use of dose coefficients provided in package inserts (PI), as
illustrated above. Most PIs were initially issued at the time of approval of a radiopharmaceutical, and dosimetry
information included in subsequent revisions has not been updated to reflect new biokinetic data or changes
in the ICRP dosimetry system. Note that in some cases there can also be substantial discordance in dose estimates between different manufactures PIs (e.g., Tl-201). In general, the most recent ICRP data should be used
for dosimetry estimates. PI should only be used as the source of dose information for new radiopharmaceuticals for which ICRP dosimetry data have not yet been published. (From Einstein AJ, Moser KW, Thompson RC,
et al. Radiation dose to patients from cardiac diagnostic imaging. Circulation 2007;116:12901305.)
1002
TABLE H-2DOSES FROM SEVERAL SOURCES OF PUBLIC AND
OCCUPATIONAL RADIATION EXPOSURE

TYPICAL EFFECTIVE
DOSE (mSv)
RADIATION EXPOSURE
US annual average per capita dose from natural background

radiation
3.1
US annual average per capita dose from medical radiation exposure
Total US annual average per capita dose from all sources
6.2
Commercial aviation aircrew average annual dose from cosmic

radiation
Radiation workers annual average occupational dose
25
Whole body x-ray airport scanner (1 scan)
0.000030.0001
Flying on airplane at 35,000 ft (per minute)
0.00004
1. NCRP. National Council on Radiation Protection and Measurements. Ionizing Radiation Exposure of the
Population of the United States, NCRP Report No. 160 National Council on Radiation Protection and
Measurements, Bethesda, Maryland, 2009.
2. Commentary No. 16Screening of Humans for Security Purposes Using Ionizing Radiation Scanning
Systems. National Council on Radiation Protection & Measurements. Retrieved November 12, 2010.
TABLE H-3 ESTIMATED LOSS OF LIFE EXPECTANCY FROM HEALTH RISKS

HEALTH risk
ESTIMATES OF AVERAGE DAYS (Years)

OF LIFE EXPECTANCY LOST
Smoking 20 cigarettes/d
2,250 (6.2 years)
All accidents combined
366 (1 year)
Auto accidents
207
Alcohol consumption
125
Home accidents
74
Drowning
24
Natural background radiation
30
Medical diagnostic x-rays
30
With the exception of smoking, all risks are expressed as the loss of life expectancy averaged over the U.S.
population.
Data from Cohen BL. Catalog of risks extended and updated. Health Phys 1991;61(3):317335. It is important
to note that there are a lot of assumptions that are incorporated into the numerical values presented in this
table. The original text should be consulted for the details and caution should be exercised when comparing a
risk from one source or activity to another (see: Covello, VT. Risk comparisons and risk communication: Issues
and problems in comparing health and environmental risks. In Kasperson RE & Stallen PJM. (eds.), Communicating risks to the public: International perspectives. London, UK: Kluwer Academic. 1991:79124. and Risk
Communication, Risk Statistics, and Risk Comparisons: A Manual for Plant Managers by Vincent T. Covello, Peter M. Sandman, and Paul Slovic Washington, DC: Chemical Manufacturers Association, 1988 Appendix B Risk
Comparison Tables and Figuresby Vincent T. Covello, Peter M. Sandman, and Paul Slovic accessed at http://
www.psandman.com/articles/cma-appb.htm# on August 1, 2011.
Appendix H Radiation Dose: Perspectives and Comparisons
1003
TABLE H-4 RADIOACTIVITY OF SOME NATURAL AND OTHER MATERIALS

1 kg of granite
1,000 Bq
0.027 mCi
1 kg of coffee
1,000 Bq
0.03 mCi
1 kg of coal ash
2,000 Bq
0.054 mCi
The air in a 100 m Australian home (radon)
3,000 Bq
0.08 mCi
1 kg superphosphate fertilizer
5,000 Bq
0.14 mCi
1 adult human (100 Bq/kg)
7,000 Bq
0.19 mCi
The air in many 100 m2 European homes (radon)
up to 30,000 Bq
0.81 mCi
1 household smoke detector (with americium)
37 000 Bq
1 kg uranium
25 million Bq
1.0 mCi
676 mCi
Radioisotope for medical diagnosis
370 MBq
10 mCi
1 luminous exit sign (1970s)
1,000,000 million
Bq (1 TBq)
27 Ci
Radioisotope source for medical radiotherapy

(Gamma Knife)
222,222,222 million
Bq (222 TBq)
6,000 Ci
Adapted from Hall E. Radiation and Life. 2nd ed. Pergamon Press, New York, 1984.
The annual outdoor effective dose (mSv)

from cosmic radiation for Canada and the U.S.
Figure H-4 Color plot of the annual outdoor effective dose from cosmic radiation (in microsievert) in
North America. (Source: Grasty RL, Lamarre JR. The annual effective dose from natural sources of ionizing
radiation in Canada. Radiat Protect Dosim 2004;108(3):215226.)
1004
Average Annual Effective Doses from Natural Radiation Sources
8
mSv
Cosmic rays
Gamma outdoors
Gamma indoors
United States
UK
Australia
Switzerland
Spain
Sweden
Norway
Portugal
Netherlands
Luxembourg
Italy
Ireland
Greece
France
Germany
Finland
Denmark
Austria
Belgium
Radon
Figure H-5 Examples of variation in natural background and the sources of the variation. (Adapted from Green
BMR, Hughes JS, Lomas PR, et al. Natural radiation atlas of Europe. Radiat Protect Dosim 1992;45:491493.)
APPENDIX
Radionuclide Therapy Home

Care Guidelines
I.1 General Safety Guide for Outpatients Receiving
Radioiodine Therapy: Less Than 10 mci
The radioactive iodine that you have been treated with is for your benefit. You need to
take some precautions so that others do not receive unnecessary radiation e xposure.
A. Please follow the instructions below:
1. Breast feeding your infant must be discontinued.
2. Drink plenty of liquids.
3. Women of child bearing potential must have a negative pregnancy test prior to
therapy. Results ______________.
B. Please follow the steps below during the first 2 days after your treatment.
1. Most of the remaining radioiodine in your body will come out through the
urine. A small portion of radioactivity will be found in your saliva and sweat.
To decrease the spread of radioactivity:
a. Use paper drinking cups, plates and plastic utensils. Put these items in the
trash when you are done with them.
b. Use towels and washcloth that only you will touch.
c. You should sleep in a separate bed.
d. At the end of 2 days wash your laundry separately. This need only be done
once.
e. Avoid touching and hugging babies and pregnant women.
2. Most of the radioactivity is in the urine during these first 2 days. If urine
should be spilled or splashed, wash and rinse the affected area three times,
using paper towels or tissue. Be sure to carefully wash your hands after using
the bathroom.
3. Breast-Feeding: Breast-feeding must be discontinued. Your physician will
advise you when you may resume.
C. During the rest of the week
Your body will still contain some radioactivity. Avoid sitting close to others for
hours at a time. This will reduce their radiation exposure. Do not be concerned
about being close to people for a few minutes. Avoid holding babies or young
children for a long time each day.
If you have problems or questions about the above, please call:
______________________, M.D.
Nuclear Medicine Division Ph. No.___________________
1005
1006
I.2 General Safety Guide for Outpatients Receiving
Radioiodine Therapy: More Than 10 mci
The radioactive iodine that you have been treated with is for your benefit. You need to
take some precautions so that others do not receive unnecessary radiation exposure.
A. Please follow the instructions below:
Young people are more sensitive to radiation. To minimize possible effects from
radiation, do not hold children or spend time near pregnant women during the
next week.
1. Go directly home. Do not stay at a hotel.
2. Sit as far from anyone as practical during your ride home from the hospital.
Let us know if your trip is longer than 3 hours.
3. Breast feeding your infant must be discontinued.
4. Drink plenty of liquids.
5. Women of child bearing potential must have a negative pregnancy test prior to
therapy. Results ______________.
B. Please follow the steps below during the first 4 days after your treatment.
1. Most of the radioiodine in your body will come out in the urine and stool.
A small portion of radioactivity will be found in your saliva and sweat. To
decrease the spread of radioactivity:
a. If possible, use a separate bath room.
b. Flush the toilet two times after each use.
c. Men should sit down when urinating.
d. If urine is spilled/splashed, wash and rinse the spill area three times, using
paper towels or tissue.
e. Menstruating women should use tampons that can be flushed down the
toilet.
f. Be sure to carefully wash your hands after using the bathroom.
g. Do not share utensils or food with others. (For example do not drink from
the same glass or share a sandwich.)
h. Use towels and washcloths that only you will touch.
i. Run water in the sink while brushing your teeth or shaving and rinse the
sink well after use.
j. Nausea following therapy is very rare. However, if you feel sick to your
stomach, try to vomit into the toilet. If necessary, wash and rinse any spill
areas three times, using paper towel/tissue.
k. At the end of 4 days wash your laundry, including the pillow cases,
separately from others. This need only be done once.
2. To decrease the radiation exposure to others:
a. You should sleep in a separate bed. Cover the pillow with two pillow cases
or a water resistant cover if possible.
b. Remain in your home for the first 4 days.
c. Young people are more sensitive to radiation so do not hold children or
spend time near pregnant women.
d. Family members should stay six feet or more from you. After the first 2
days, they may be closer for brief periods such as a few minutes.
e. Avoid sexual relations.
Appendix I Radionuclide Therapy Home Care Guidelines
1007
C. Please follow the steps below until the end of the first week after your
treatment.
Your body will still contain some radioactivity. To minimize radiation exposure to
other people, you should:
1. Avoid sitting close (within a foot) to others for hours at a time.
2. Minimize use of public transportation (bus, airplane, etc.) and going to public
gatherings such as movies, plays, etc. Do not be concerned about being close
to people for short times (less than an hour).
3. To minimize possible effects from radiation, avoid holding babies or children.
D. Please follow the steps below for the next 6 to 12 months
1. Avoid becoming pregnant for the next 12 months or fathering a child for the
next 6 months.
2. If you are hospitalized or require emergency medical treatment within the
next 2 weeks, please inform the doctor or nurse that you have been treated
with ________ mCi of radioactive iodine on _________. Please contact the
Nuclear Medicine Department at __________________ as soon as possible.
After hours, call the hospital operator at _____________ and ask for the
Nuclear Medicine on-call physician.
________________________, M.D. _______________________, M.D.
Nuclear Medicine Physician SignatureNuclear Medicine
Printed Name
Date: ______________________________
Physician
I understand the above instructions which have been discussed with me and
I agree to follow them.
______________________________ ____________________________
Patients Signature Date
Index
(Note: Page numbers followed by f indicate figures; page numbers followed by t indicate tables)
A
ABC. See Automatic exposure rate control
Abscopal effect, radiation, 771
Absolute risk, radiation biology, 795
Absorbed dose, 376, 386387
radiopharmaceuticals, 978t979t
x-ray, 953t955t
Absorption
efficiency, 225226
radiation, 18
Acceleration, 915
Access, random and sequential, 113
Accident, radiation protection, 902903
Accuracy
ROC curve, 98
Acoustic impedance, ultrasound, 507, 507t
Acoustic power, pulsed ultrasound intensity,
572574, 573f, 573t
Acquisition
digital images, 145146, 146f
MR parameters
echo time, 420421
inversion time, 421
partial saturation, 421, 421f
repetition time, 420
Active transport, radiopharmaceuticals, 614
Activity, radionuclide decay, defined, 579, 580t
Acute radiation syndrome (ARS), 785791
gastrointestinal syndrome, 788790
hematopoietic syndrome, 786788, 788f, 789f
neurovascular syndrome, 790
sequence of events, 782783, 783f, 784t, 785786,
786f
treatment, 906907
Added filtration, x-ray, 189
Advisory bodies, radiation protection, 893
Air gap
anti-scatter grid, 235236, 235f
scattered radiation, mammography, 256
Air kerma, 55
ALARA. See As low as reasonably achievable
Aliasing, image quality and, 553, 553f
Alpha decay, nuclear transformation, 583584, 583f
Alternating (AC) electric current, 180
defined, 920921, 920f
Ampere, defined, 918
Amplitude, electromagnetic wave, 21
Analog-to-digital converters (ADCs), 107
Anatomical noise, 78f, 79
Anger camera. See Scintillation camera
Angiography
CT, 344345, 346f
MR angiography, 465
phase contrast, 467469, 468f, 469f
time-of-flight, 465466, 466f, 467f
Annihilation coincidence detection (ACD). See
Positron emission tomography
Annihilation photons, positron annihilation, 37,
722723
Annihilation radiation, 12, 587, 587f
Anode, 118, 171
cooling chart, 201202, 201f
grounded, 241
mammography, 241
Anti-scatter grid, 256
air gap as alternative to, 257
Bucky factor, 234
contrast degradation factor, 235
focal length, 233234
grid frequency, 233
grid ratio, 233
grid type, 233
interspace material, 233
moving grids, 234
primary transmission factor, 234235
scatter transmission factor, 235
selectivity, 235
Apoptosis, DNA damage, 759
Artifacts
film processor, 212
magnetic resonance imaging, 474486
PET/CT imaging
attenuation correction artifacts, 740, 740f
spatial misregistration, 739, 739f
truncation artifacts, 741
ultrasound, 560568
x-ray computed tomography, 367370
As low as reasonably achievable (ALARA),
895, 897
Atom
excited state, 28
fusion, 32, 32f
isomeric states, 28
structure, 2432
atomic binding energy, 31
atomic nucleus, 2730
atomic number, 27
binding energy, 31
electron binding energy, 2425, 25f
electron cascade, 26
1009
1010
Index
Atom (Continued)
electron orbit, 2426, 27f
electronic structure, 2426
energy level, 25f
fission, 31, 32f
ground state, 2728
mass defect, 31
mass number, 27
quantum numbers, electron shell, 24
valence shell, 24
Atomic binding energy, 31, 32f
Atomic number, 27
Attenuation
electromagnetic radiation
effective energy, 4950
half value layer, 4849, 50t
homogeneity coefficient, 51
linear attenuation coefficient, 4546, 47t
mass attenuation coefficient, 4748, 48f
mean free path, 5051
mass attenuation coefficients, 946t954t
ultrasound, 511513, 511t, 512f
x-ray dosimetry, 375376
Auger electron, 2627, 27f
Automatic exposure rate control (AERC), 292, 293f
Average glandular dose, radiation dosimetry,
mammography, 274
Axial resolution, ultrasound, 524525, 525f
B
Backprojection
single photon emission computed tomography
(SPECT), 711, 711f
x-ray computed tomography, 352354, 353f,
354f, 355f
Backscatter factor, 380
Backup timer, 252
Bandwidth, 134
Magnetic Resonance RF, 461462
network, 134
ultrasound, 517518
Beam hardening, 51, 51f
x-ray computed tomography, 367, 367f, 368f
Beta particles, 22
Beta-minus decay, 584585, 584f, 585f
Beta-plus decay, 585586, 586f
Binary number, 101103, 102t
Binary energy
electron, 2425
nuclear, 31, 32f
Binomial distribution, 81, 669
Biplane angiographic systems, 303
Bismuth germanate (BGO), inorganic crystalline
scintillators in radiology, 645
Bit, 105
Blood oxygen level-dependent (BOLD), 470471,
471f
Blurring, physical mechanisms of spatial resolution,
62, 68
Bone kernels, x-ray computed tomography, 360, 360f
Bone mineral densitometry x-ray system, radiation
protection, 859
Bounce point, short tau inversion recovery, magnetic

resonance imaging, 429
Bow-tie filters, computed tomography, 190
Bragg peak, 35
Breast cancer, radiation induced carcinogenesis,
819821, 820t
Breast-feeding, recommended duration of interruption
of, 888, 889t
Bremsstrahlung spectrum
defined, 37
filtered, 173, 173f
process, x-ray production, 171174, 172f, 173f
radiation, 37
unfiltered, 173, 173f
Brightness gain, fluoroscopy, 289290
Broad-beam geometry, 48, 48f
Bucky factor, 234
mammography, 256257
Bucky grid, 234
Bystander effect, 771772
Bytes, 105
C
Calcium tungstate, inorganic crystalline scintillators in
radiology, 638
Cancer development, 794
Capacitive Micromachined Ultrasonic Transducer
(CMUT), 519
Cardiac catheterization suite, 302f, 303
C-arm CT, 297298, 297f
Cassette, in film screen radiography, 258263, 259f
Catalyst, developer, film processing, 211
Cataracts, radiation, 782783
Cathode, 118, 171, 176179, 178f, 179f
mammography, 240241
Cathode ray tube (CRT), 117119, 118119,
118f119f
Cellular radiobiology, 764772
adaptive response, 770771
bystander effect, 771772
cell survival curves, 764765, 764f, 766f
cellular radiosensitivity, 766770, 770t
genomic instability, 772
Center of rotation, emission tomography, 717718,
718f
Central processing unit, 113, 932934
Cesium-137, spectrum of, 655657, 656f
Chain reactions, 599602, 599f601f
Characteristic curve of film, 212214
Characteristic x-ray, 26
Charge-coupled device (CCD)
amount of light, 218
dexel, 217
slot-scan X-ray systems, 219
x-ray quantum limited detector, 219
Chemical shift artifacts, magnetic resonance imaging,
480483, 481f483f
Chemotaxis, radiopharmaceuticals, 615
Chromatid aberrations, cellular radiobiology, 759
Chromosome aberrations, cellular radiobiology, 759,
762763, 763f
Cine-radiography cameras, 303, 309

Classical scattering. See Rayleigh scattering
Clustered damage, DNA, 754
Code of Federal Regulations (CFR), 629, 892
Coherent scattering. See Rayleigh scattering
Coil quality factor, magnetic resonance imaging,
462473, 463f
Collective effective dose, 840, 840t
Collimator, 171
axial, 727
converging, 678f, 679
diverging, 678f, 679
efficiency, 688690, 689f, 690t
fan-beam, 679, 714, 714f
mammography, 250
nuclear imaging, 677679, 678f
parallel-hole, 678, 678f
pinhole, 678f, 679
resolution, 683
scintillation camera, 677679, 678f
x-ray, 187, 188f
Collimator resolution, performance, scintillation
camera, 683, 688690, 689f, 690t
Committed dose equivalent, radiation protection, 894
Complementary metal-oxide semiconductor (CMOS),
219220
Compression, mammography, 253255
Compton scattering, 20, 3941, 40f
Computed tomography (CT), 67, 7f, 955, 957t
acquisition modes
axial/sequential acquisition, 335337, 336f, 337f
cardiac CT, 340343, 341f, 342f, 343f
cone beam acquisition, 340, 340f
CT angiography, 344345, 346f
CT perfusion imaging, 345346, 347f
dose modulation, 347350, 348f, 349f
dual energy CT, 344, 345f
helical (spiral) acquisition, 337340, 337f,
338f, 339f
scanned projection radiograph, 335, 336f
shuttle mode, 346347, 347f
basic principles, 318319
beam hardening, 367, 367f, 368f
cone beam artifacts, 369370, 370f
cone beam reconstruction, 356
detector type, 329335
fan beam geometry, 316
filtered backprojection, 352354, 353f, 354f, 355f
Fourier-based reconstruction, 354356, 356f
Hounsfield units, 324
iterative reconstruction, 357358, 357f, 358f
partial volume, 368369, 369f
patient protection, radiation protection, 874877,
875t, 876t
preprocessing, 350351, 351f
protocols, 350
radiation dose, 953955t
scanner, barrier, 866
scanner, shielding, 864865, 865f, 866f
simple backprojection, 351352, 352f, 353f
slip-ring technology
streak artifacts, 367368, 368f
view aliasing, 368, 369f
Index
1011
Computed tomography dose index (CTDI), 387394

AAPM Report 204, 390
definition, 389
function of time, 393, 393f
helical pitch, 389
ionization chamber, 388
length of the phantom, 392393, 393f
PMMA phantom, 388, 388f
rise to dose equilibrium curve, 392393, 392f
scattered dose distribution, 390391
velocity, 393, 393f
Computer network, 133142
Computer-aided detection (also diagnosis), 160
Computer-assisted detection, mammography, 270
Computers, 111112
components and operation, 929930
acquisition interface, 934935
communications interfaces, 935
computer program and central processing unit,
932934
display interface, 934
expansion slots, 934
input/output bus, 934
main memory, 931
mass storage devices, 934
specialized processors, 935, 936f
performance, 935936
software, 937
Computers in nuclear imaging, 698704
Conditional radiosensitivities, 766769, 767f768f
Conformance statement, DICOM, 148
Constant potential x-ray generator, 197
Constructive/destructive wave interference
ultrasound, 504, 505t
Containment structure, radionuclides, 601, 602
Contamination control, radiation protection, 883, 885
Contrast
computer display, 126128, 126f, 127f
computer enhancement
translation table, 126, 127f
windowing, 127128
degradation factor, 235
ultrasound, 562
Contrast ratio, 122
Contrast resolution, 300
Contrast-to-noise ratio (CNR), 91, 91f
Control rods, 600
Converging collimator, 678f, 679
Conversion
efficiency, 225226
factors, definitions, 938t940t
Convolution, 6568
and spatial resolution, 994
Counting statistics, 667673
data characterization, 667668
estimated standard deviation, 670
probability distribution functions, for binary
processes, 668, 668t
propagation of error, 671673, 672t
propagation of error probability distribution
functions, 671673, 672t
1012
Index
Counting statistics (Continued)

random error, 667
sources of error, 667
systematic error, 667
CPU. See Central processing unit
Critical, nuclear reactor, 600
Cross-excitation, magnetic resonance imaging,
452, 463
CT dosimetry
CTDI
AAPM Report#No 204, 390
definition, 389
helical pitch, 389
ionization chamber, 388
length of the phantom, 392393, 393f
PMMA phantom, 388, 388f
rise to dose equilibrium curve, 392393, 392f
scattered dose distribution, 390391
velocity, 393, 393f
dose in, 394
organ doses, 387
CT number. See Hounsfield unit
Current mode, radiation detection, 364365,
634635, 635f
Cutaneous radiation syndrome, 775
Cyclotron-produced radionuclide, 594598, 595f, 597f
D
DAC. See Digital-to-analog converter
Data conversion
analog-to-digital conversion, 107109, 108t
digital-to-analog conversion, 109, 109f
digitization, 108
quantization, 108
sampling, 108
signal-to-noise ratio, 109
Decay constant, radionuclide decay defined, 579580
Decay equation, radionuclide decay, 582, 583f
Decay schemes
fluorine-18, 581t, 592f
molybdenum-99, 580, 581t, 591f
nuclear transformation, 589593, 589f592f, 593t
phosphorus-32, 581t, 589, 590f
radon-220, 589, 590f
technetium-99m, 580, 581t, 592f
Decibel
scale, and relative intensity values, 505, 506t
sound, 506, 506t
Decision criterion, receiver operating characteristic
curves, 96, 96f
Decision matrix, 2 x 2, image quality, 96
Deconvolution, and spatial resolution, 67
Delta function, and spatial resolution, 67
Delta ray, particle, 33
Densitometer, 261, 261f
Design goals, radiation protection, 855, 856
Detection efficiency, radiation detection, 85,
636637, 637f
Detective quantum efficiency (DQE), 9495, 95f
Detector aperture width, image quality, 73f
Detector contrast, and image quality, 8889, 89f
Detector pitch, 75
Deterministic effect. See Tissue reactions

Deviation index (DI), 228
Diamagnetic materials, 927928
DICOM. See Digital Imaging and Communications in
Medicine
Diffusion contrast, 469473
Digital full-field mammography, 263268
Digital images
acquisition of
film digitizer, 145146, 146f
image formats, 124t, 145
display of
cameras, 133
diagnostic monitors, 132133
display interface and image conversion, 117
display monitors, 117125
false color images, 131132, 132f
image contrast and contrast enhancement,
126128
luminance calibration, 128131
monitor pixel formats, 125126
Digital Imaging and Communications in Medicine
(DICOM), 147148
Digital image processing
computer-aided detection, 160
image co-registration, 160
spatial filtering, 159
volumetric image data
maximum intensity projection, 161, 162f
multiplanar reconstruction, 160161, 162f
stack versus tile mode display, 160
volume rendering, 161, 162, 162f
Digital mammography, 6
charge- coupled-device (CCD)-detector array, 267
direct conversion, 265
full-field digital mammography, 263268
image acquisition, 272, 273f
indirect conversion, 265
processing and viewing, 268270
screening exams, 267t
Digital radiological images, 109111, 110f, 111f
Digital-to-analog converter (DAC), 108109, 109f
Digital tomosynthesis, mammography,
272273, 273f
Diode, x-ray, 193, 633
Direct current (DC), 193
defined, 916917, 916f
Direct x-ray conversion TFT arrays, 222
Disk, data storage
magnetic, 113114, 113f
optical, 114115, 115t
Disk mirroring, RAID, 116
Display, computer
contrast enhancement, 126128, 126f, 127f
digital conversion to analog video signal, 117
Display of images
ambient viewing conditions, 156
calibration of display workstations
display workstations, 154156, 155f
monitor pixel formats, 125126
Displayed contrast, and image quality, 8990, 90f
Diverging collimator, 678f, 679

DNA, radiobiology, 757760, 759f761f
Doppler frequency shift, 542544, 543f, 544t
Doppler spectrum, 551
Doppler ultrasound, 11, 542554
aliasing, 553
color flow imaging, 548551, 549f, 552f,
553f, 561f
continuous Doppler operation, 544545, 545f
Doppler frequency shift, 542544, 543f, 544t
Doppler spectral interpretation, 551553,
554f, 555f
duplex scanning, 548
power Doppler, 553554, 554f
pulsed Doppler operation, 545547, 546f, 547f
quadrature detection, 545
Dose
absorbed, 5354, 376
annual cosmic radiation doses, 838, 843f
vs. effective dose, 998, 998f
effective, 385386, 972t973t
days of natural background equivalent to effective dose from imaging procedures, 952-953
newborn and infant, 978t
pediatric, 955, 958t
pediatric anthropomorphic phantom, 955, 958t
entrance skin kerma, 379380
equivalent, 383384
conceptus doses, 955, 959t
organ-specific radiation doses, 955, 957t
fluoroscopy
dose reduction, 309, 310t
dose to personnel, 309
patient dose monitoring, 306, 308309
patient dose rate, 304305, 304f
personnel, 305306
recording and patient follow-up, 310311
limit, 894
mammography, 272, 273f
natural background doses, 838,843f
radiation. See Radiation dose
estimated effective doses vs.standard myocardial
perfusion, 998, 1001f
ICRP tissue weighting factors, 998, 999
life expectancy, 998, 1002t
natural and artificially produced sources, 998, 1003t
organ equivalent doses, 998, 1000f
public and occupational radiation exposure,
998, 1002t
radiopharmaceuticals
absorbed, 981t982t
adult dose, 973t976t
breast dose, 986t
skin, 380382
source-to-skin distance, 377378, 379f
tube output, 377, 378f, 378t
Dose and dose-rate effectiveness factor (DDREF), 797
Dose calibrator
for assaying radiopharmaceuticals, 666
molybdenum concentration, 666
quality assurance, 665666, 665f
Index
1013
Dosimeter. See Personnel dosimetry

Doubling dose, estimating genetic risk, 822
Dual modality imaging
PET/CT imaging
artifacts
attenuation correction, 737738, 738t
PET/MRI system, 741742
quality control, 742
SPECT/X-ray CT system, 741
Dual-energy radiography
chest radiography, 229
compton scattering component, 229
logarithmic weighted subtraction, 230
Duplex scanning, Doppler ultrasound, 548
Dynamic aperture, ultrasound beam, 523
Dynamic, frame mode acquisition, image
acquisition, 697
E
Early transient erythema, 776
Echo planar image (EPI) acquisition, 454456, 455f
Eddy currents, 439
Edge gradient, 64
Edge spread function (ESF), 6465, 65f
Effective anode angle, 241
Effective dose, 385386
adult
CT procedures, 955, 957t
dental, 955, 957t
diagnostic radiology, 955, 955t
interventional radiology procedures, 955, 957t
cautionary notes, 385386
days of natural background equivalent to effective
dose from imaging procedures, 952953
definition 57, 58t
k factors, 390
organ-specific radiation doses, 955, 957t
pediatric, 955, 958t
pediatric anthropomorphic phantom, 955, 958t
per capita, 840f
radiopharmaceuticals, 977t978t
Effective dose equivalent, 894
Effective focal spot length, 183, 183f
Efficiency
conversion efficiency, 643, 646t
detection efficiency, 636637, 637f
geometric, 636
intrinsic, 636
radiation detector, 633
Electrical current, 190191, 191f
Electrical dipole, 513, 514f
Electrical potential difference (voltage), 190
Electrical power, 114, 191, 562
Electricity
conductors, 921922
direct and alternating current, 920921
electrical charge, 917
electrical current, 918, 919f
1014
Index
Electricity (Continued)
electrical dipole, 917
electrical forces and fields, 917
electrical potential difference, 918919
electrical power, 920
insulators and semiconductors, 921922
Electromagnetic induction, 190191, 191f, 920
Electromagnetic radiation
attenuation
beam hardening, 51, 51f
effective energy, 4950
half value layer, 4849, 50t
homogeneity coefficient, 51
linear attenuation coefficient, 4546, 47t
mass attenuation coefficient, 4748, 48f, 946t954t
mean free path, 5051
interactions
Compton scattering, 3941, 40f
pair production, 44, 45f
photoelectric absorption
Rayleigh scattering, 3839, 39f
particle characteristics, 22
Electromagnetic spectrum, 19f
Electromagnetic wave characteristics, 21, 21f, 22f
Electron
anode, target of, 179180
binding energy, 2425, 25f
capture decay, nuclear transformation, 587588, 588f
cascade, 26
cathode, source of, 177179, 178f, 179f
elastic interaction, 36
inelastic interaction, 37
interactions, 37
orbit, 2425, 25f
shell, 2425, 25f
transition
Auger electron, 2627, 27f
characteristic x-ray, 26
electron cascade, 26
fluorescent yield, 27
radiation from, 2627, 27f
valence shell, 24
Electron optics, fluoroscopy, 285286, 285f, 286f
Electronic gain, fluoroscopy, 285
Electronic noise, 78
Electronic structure
atom, 2426, 25f
binding energy, 2425, 25f
Electronic switch/timer, exposure timing, 198
Elevational resolution, ultrasound, 524f, 527, 527f
Emission imaging, 680, 681
Emission, x-ray, factors affecting, 202206
Energy
conservation, 916
kinetic, 916
potential, 916
Energy absorption efficiency, 225
Energy fluence, 52
Energy resolution
definition, 658659, 659f
scintillation camera, 686
Enhancement artifact, ultrasound, 527f, 529
Entrance skin kerma, 379380

Erythema, 7512,775778, 779f
Equivalent dose
definition, 5657, 58t
organ dose in x-ray imaging 383384, 954t
Ethernet, 136138, 137t
Excitation
particle, 3337, 34f, 633
Excited state, 28
Exposure
definition, 5455
measurement of, 642
timing, x-ray, 176177
Exposure control, 852854
distance, 853854, 853f
principles of, 852
shielding, 854
time, 852
Exposure index, 226228
Exposure rate constant, radiation protection, 881, 883t
Exposure time, nuclear medicine, 880, 880t
F
F-18 Fluorodeoxyglucose (F-18 FDG)
clinical utility, 742743
standardized uptake value, 745
False-positive fraction (FPF), 97
Fan beam geometry, x-ray computed
tomography, 316
Fan-beam collimator, 679, 690, 691f, 714, 714f
Far field, ultrasound beam, 521
Fast neutron, nuclear reactor produced
radionuclide,599
Fast spin echo (FSE) techniques, 453454, 453f
Ferromagnetic materials, 928
Fetal growth stage, 824, 825, 826f
Fetus
dose from radiopharmaceuticals, 627, 628
dose limits, 894, 896t
radiation effects, in utero, 823832
calculation of conceptus dose, 832, 836t
Field of view
magnetic resonance imaging 405, 439444
nuclear medicine collimators, 679
spatial resolution, x-ray computed tomography,
294296
Filament circuit, x-ray, 177, 178f
Film badge, 843844, 844f
Film digitizers, 145146, 146f
Film processing, 211212, 259260
screen-film cassettes, 258263
screen-film radiography, 212213
Film sensitometry and densitometry, mammography,
260262
Filtered backprojection reconstruction
Fourier transform, 990991
Filtered backprojection reconstruction
PET, 730
SPECT, 709711, 711f
x-ray computed tomography, 352354, 353f,
354f, 355f
Filtered Bremsstrahlung spectrum, 173, 173f

Filtration, x-ray, 189190
Fission, 14
Fission-produced radionuclides, 602
Flash memory, 114
Flat panel detector digital fluoroscopy systems,
290291, 291f
Flat panel thin-film-transistor (TFT), 220222
composition, 220
direct X-ray conversion, 222
gate and drain lines, 221222
indirect x-ray conversion, 221222
Flicker
display monitors, 123
fluoroscopy, 289
Flip angle, magnetic resonance imaging, 414415, 415f
Flip angles, 414415, 415f
Fluence, 52
Fluorescent yield, 27
Fluorine-18, decay scheme, 581t, 592f
Fluoroscopic imaging systems, radiation protection, 857
Fluoroscopy, 5
automatic exposure rate control, 292, 293f
biplane angiographic systems, 303
cardiology catheterization suite, 302f, 303
cardiology electrophysiology laboratory, 303
chain components, 283284, 283f
detector systems
camera systems, 290
electron optics, 285286, 285f, 286f
image intensifier, 284, 284f
input screen, 284285, 285f
optical coupling, 287
optics, 285286
output phosphor, 286287, 286f
video cameras, 288289, 288f
dose-related metrics, 377382
functionality, 282
genitourinary, 301
image quality
contrast resolution, 300
spatial resolution, 298299, 299f
temporal resolution, 300301
mobile, 303
modes of operation
C-arm CT, 297298, 297f
continuous fluoroscopy, 293
field of view, 294296, 295f
frame averaging, 296297, 296f
last-frame-hold, 297
last-sequence-display, 297
magnification modes, 294296
pulsed fluoroscopy, 293294, 294f
road mapping, 297
radiation dose
dose monitoring, 306311
dose to personnel, 305306
patient dose rate, 304305
radiation protection, 872874
remote fluoroscopy rooms, 301302
shielding, 857
skin dose, 380382
Flux, 52
Index
1015
F-number, lenses, fluoroscopy, 287

Focal plane tomography, 705706
Focal spot
effect on spatial resolution, x-ray computed
tomography, 366
mammography, 241244
size, anode, 182184
Focusing cup, cathode, x-ray tubes, 177179, 179f
Forces, 915
Fourier analysis, 992993, 992f
Fourier transform, 6970
filtered backprojection reconstruction, 990991
gaussian, 992993, 992f
modulation transfer function, 991995, 992f994f
RECT, 993994, 993f
Fraunhofer region, ultrasound beam, 521
Free radical, 754756, 769,776
Frequency
electromagnetic wave characteristics, 21
sound, 501f, 502504, 503t, 505f
Frequency domain, 69
Frequency encode gradient, magnetic resonance
imaging, 441442, 442f
Fresnel region, ultrasound beam, 520
Fuel rods, 600
Fusion, image, 1415
G
Gamma camera. See scintillation camera
Gamma ray, 20, 30
Gas-filled radiation detector, 637643, 640f, 642f
Gastrointestinal radiation syndrome, 788790
Gated, frame mode acquisition, image acquisition,
699700, 700f
Gauss, 403
Gaussian distribution, image quality, 669, 670f
GeigerMueller radiation counter, 639, 641643, 642f
Genetic effect, radiation. See Hereditary effects
Geometric efficiency, radiation detection, 636
Geometry: area, surface, volume, circumference,
938t939t
Gibbs phenomenon, magnetic resonance imaging, 483
Gonadal shielding, radiation protection, 871
Gradient coils, 405
Gradient field strength, 439
Gradient moment nulling, 469, 469f
Gradient recalled echo, gradient echo, 434f, 434t, 435
Gradient sequencing, magnetic resonance imaging, 444
Grain noise, image quality, 77, 78f
Grain noise, image quality, 77, 78f
Grating lobe artifact, ultrasound, 523f, 524f, 565f,
566567
Gray scale value, 15, 60, 75, 90f, 106, 158f
Gray, unit of absorbed dose, 53
Grid frequency, 233
Grounded anodes, 241
Gyromagnetic ratio, 410
H
Half-life
biological, 622, 622f, 905
effective, 609, 620, 621
physical, 580581, 581t
1016
Index
Half-value layer (HVL), 4849, 50t, 378t

mammography, 247248
Harmonic imaging, 518, 556559, 557f, 558f
Health Insurance Portability and Accountability Act
(HIPAA), 157, 163
Heat loading, 200202, 201f
Heel effect, anode, 184185, 186f
Helical computed tomography, 337340, 337f,
338f, 339f
Hematopoietic radiation syndrome, 786788, 788f, 789f
Hereditary effects of radiation exposure, 821823
Hertz, unit, 21
Homogeneity coefficient, 51
Hospital information system (HIS), 144, 148149
Hot spots, radiopharmaceuticals, 613
Hounsfield unit, x-ray computed tomography, 324
Huygens principle, ultrasound beam, 520
I
Illuminance, film viewing conditions,
mammography, 263
Image format, digital images, 698
Image intensifier systems
characteristics
brightness gain, 289290
pincushion distortion, 290
fluoroscopy, 284, 284f
Image plates, computed radiography, 214, 215f
Image quality
and aliasing, 73, 73f, 74
contrast, 8790, 88f, 89f, 90f
contrast resolution, 7686
contrast-detail curve, 9294, 93f, 94f
contrast-to-noise ratio, 91, 91f
convolution, 6568
detective quantum efficiency, 9495, 95f
detector contrast, 8889, 89f
digital image contrast, 89, 90f
displayed contrast, 8990, 90f
frequency domain, 69
Fourier series and the Fourier transform, 6970
modulation transfer function, 7076
image noise source
anatomical noise, 78f, 79
electronic noise, 78
grain noise, 77, 78f
nonparametric measures, 8081, 80f
normal distribution, 8184, 81f, 82f, 83f
Poisson distribution, 8485, 84f, 86t
quantum noise, 80
statistical distributions, 80
structure noise, 78f, 79
noise texture, 8687, 86f, 87f
receiver operating characteristic (ROC) curves,
9699, 96f, 97f, 99f
signal-to-noise ratio, 9192, 92f
spatial resolution
edge spread function, 64, 65f
line spread function, 6364, 63f, 64f
point spread function, 6163, 61f, 62f
spatial domain, 6061
subject contrast, 8788, 88f
Image reconstruction
computed tomography
SPECT and PET, 709711, 710f, 711f, 712f, 730
Imaging, medical modalities
functional, 11
projection, 5
transmission, 5
Imparted energy, 5556
In utero radiation effects, 806, 823833
Indirect biologic changes, radiation interaction, 754
Indirect detection process, 211f
Indirect X-ray conversion TFT arrays, 221, 222f
Inherent infiltration, x-ray, 189
Initiation, cancer development stages, 794
Input screen, fluoroscopy, 284285, 285f
Integrating the Healthcare Enterprise (IHE), 149
Intensifying screens, radiography, 210, 224225
Intensity, ultrasound, 501f, 504506, 506t
Internal conversion electron, 30
Internal dosimetry
body dose and effective dose, 626628
dose factors, 624t, 625t
MIRD formalism, 617618, 618f
RADAR formalism, 618620, 619f
radiopharmaceutical, 617
time-integrated activity
compartmental model, 620, 621f
dose calculation, 623, 624t, 625t
mean life, 621
simple kinetic model, 620, 621f
International Commission on Radiological Protection
(ICRP), 893
dose definitions and dose weighting factors,
5259
principles for radiation protection, 852
International System- of Units, 914, 915t
Internet, 139
Internet Protocol Suite, 139
Interventional facilities, radiation protection, 857
Intrinsic efficiency, radiation detection, 637
Intrinsic resolution, performance, scintillation
camera, 683
Inverse square law, 236, 249, 258, 853, 853f
Inversion recovery, magnetic resonance
imaging, 464
Inversion recovery spin echo (IR SE), 427
Inverter x-ray generator, 195, 195f
Iodine-125, spectrum of, 657658, 658f
Ion pair, 33
Ionization, 22, 3437, 34f, 35f
chamber, gas-filled radiation detector, 637643,
640f, 642f
Ionizing radiation, 22
Isobaric transitions
beta-minus decay, 588589
nuclear transformation, 584
Isobars, 28t
Isomeric state, 28
Isomeric transition, 30
Isomers, 28t
Isotones, 28t

Isotopes, 28t
Iterative reconstruction
SPECT and PET, 711, 712f, 730
x-ray computed tomography, 357358, 357f, 358f
J
Joint Commission. See The Joint Commission
Joule
definition, 200201
rating chart, x-ray tube, 376
K
K shell, electron, 2426
K-absorption edge, 244
Kerma, 5253
Kerma-area-product (KAP), 306
Kernel, and spatial resolution, 66, 66f
computed tomography, 353354, 353f, 356f
in filtered backprojection, 710711
Keyhole filling method, magnetic resonance
imaging, 456f,
K-space data acquisition, magnetic resonance imaging
K-space errors, magnetic resonance imaging, 478, 479f
K-space filling methods, 456457
K-space matrix, 444, 445f
L
Laboratory frame, 411, 412f
Laboratory safety
radiation protection, 885887, 886f
LAN. See Local area network
Larmor equation, magnetic characteristics, 409410,
410f
Laser multiformat camera, 133
Last-frame-hold, fluoroscopy, 297
Latent image center, 211
Latent stage, acute radiation syndrome, 785, 786f
Lateral resolution, ultrasound, 525526, 526f
Latitude, radiography, 213
LCD. See Liquid crystal display
LD50/60, 787
Lead thickness, radiation shielding barriers,
859860, 858f
Leaded glasses, radiation protection, 869
Leadzirconatetitanate (PZT), ultrasound, 514
Leakage radiation, 187, 856, 856f
Leukemia, radiation induced carcinogenesis, 792t
Line focus principle, x-ray tube focal spot, 182
Line spread function (LSF), 6364, 63f, 64f, 682, 730
Linear attenuation coefficient, 4546, 47t
Linear energy transfer (LET), 36
Linear nonthreshold (LNT) model, radiation effects,
807, 807f
Linear-quadratic (LQ) model, radiation effects, 765,
807, 807f
Liquid crystal display (LCD), 119121, 120f
List-mode acquisition, scintillation camera,
700, 701f
Local area network, 136138
Logarithmic weighted subtraction, 230
Longitudinal magnetization, 412, 413f
Look-up table, displayed contrast, image quality,
89, 90f
Index
1017
Luminance, 121
film viewing, mammography, 262
digital image viewing, mammography, 269
M
Magnetic dipole, 407
Magnetic field inhomogeneities, magnetic resonance
imaging artifact, 497
Magnetic fields, 403404
Magnetic forces, 922924
Magnetic moment, defined, magnetic resonance
imaging, 412
Magnetic resonance basics
acquisition parameters
echo time, 420421
inversion time, 421
partial saturation, 421, 421f
repetition time, 420
energy absorption and emission, 410411,
410t, 411f
frame reference, 412413, 412f, 413f
geometric orientation, 412413, 412f, 413f
K-Space data acquisition
frequency domain, 444
K-space matrix, 444, 445f
Nyquist frequency, 444
two-dimensional data acquisition, 445446,
446f, 447f
two-dimensional multiplanar acquisition,
446447, 447f
magnetic characteristics
nuclear elements, 408, 408t
nucleus, 407408, 407f, 408t
proton, 409410, 410f
magnetic fields, 403404
magnetic susceptibility, 407
magnetism, 403
magnetization vectors, 412413, 412f, 413f
magnets
air core magnets, 404, 404f
bipolar gradient field, 405
electromagnet, 404
FOV, 405
gradient coils, 405, 406f
internal superconducting magnet, 405, 405f
magnetic field gradient, 405
quench, 405
RF coils, 405
shim coils, 405
superconducting air-core magnet, 405, 405f
superconductivity, 404, 405
MR signal
classical physics model, 414
electromagnetic RF wave, 414, 414f
flip angles, 414415, 415f
quantum mechanics model, 413
resonance frequency, 413
RF excitation pulse, 413
MR signal localization
frequency encode gradient, 441442,
442f, 443f
gradient sequencing, 444, 444f
1018
Index
Magnetic resonance basics (Continued)

magnetic field gradients, 438439, 438f,
439f, 439t
phase encode gradient, 443444, 443f
slice select gradient, 440441, 440f, 442f
MR system, 406, 406f
Magnetic Resonance Imaging
acquisition time, 449460
ancillary equipment, 488491
angiography, 465469, 466f469f
artifacts, 474486
biological effects, 496497
blood oxygen level-dependent (BOLD),
470471, 471f
centric k-space filling, 456457, 456f
chemical shift artifacts, 480483, 481f483f
coil quality factor, 462463
cross-excitation, 463
data synthesis, 452453, 452f
diffusion weighted imaging, 471473, 472f
echo planar image acquisition, 454456, 455f
fast spin echo (FSE) acquisition, 453454, 453f
field uniformity, 492493
flip angle, 414415, 415f
flow-related enhancement, 464465, 465f
functional MRI (fMRI), 470471, 471f
gradient field artifacts, 476477, 476f
gradient recalled echo acquisition, 454, 454f, 455t
gradient sequencing, 444445, 445f
image acquisition, 463
image quality, 463464
image reconstruction, 463
keyhole filling method, 456f, 457
K-space errors, 478, 479f
K-space filling methods, 456457
localization of signal, 422
machine-dependent artifacts, 474475
magnet siting, 491492, 492f, 493f
magnetic field
exposure limit, 497
noise limit, 497
magnetic field strength, 463
magnetic resonance spectroscopy, 486488,
487f, 488t
magnetization transfer contrast, 473474, 473f
motion artifacts, 479480, 480f
MR bioeffects and safety, 495499
acoustic noise limits, 497
guidelines, 496t
MR personnel, 498499, 498f
pediatric patient concerns, 498
pregnancy-related issues, 497
RF exposure, 497
safety zones, 498499, 498f
specific absorption rate, 497
static magnetic fields, 496497
time-varying magnetic fields, 497
multislice data acquisition, 451452, 451f
parallel imaging, 457, 458f, 459
partial volume artifacts, 486
phase encode gradient, signal localization,

455, 455f
PROPELLER, 457, 458f
quality control, 493495, 494f, 494t
radiofrequency coils, 489491, 490f
multi-channel encoding coils
phased array coils, 489490, 490f
RF receiver coils, 489
RF transmitter coils, 489
surface coils, 489
volume coils, 489
RF artifacts, 477478, 478f
RF bandwidth, 461462, 462f
RF coil artifacts, 477, 477f
RF coil quality factor, 462463
ringing artifact, 483484, 484f, 485f
signal averages, 461
signal from flow
flow-related enhancement, 464465, 465f
gradient moment nulling, 469, 469f
MR angiography, 465
phase contrast angiography, 467469, 468f, 469f
time-of-flight angiography, 465466, 466f, 467f
signal-to-noise ratio, 460461
small flip angle, gradient recalled echo, 454
spatial resolution and contrast sensitivity, 460
spin-lattice relaxation, defined, 475
spin-spin interaction
spiral filling method, 456f, 457
subcomponents, 491, 491f
susceptibility artifacts, 475476, 475f
three-dimensional Fourier transform imaging, signal
localization, 459460, 459f
two-dimensional data acquisition, 450451, 450f
voxel volume, 461
wraparound artifact, 485486, 485f
Magnetic Resonance Imaging (MRI), 89, 9f
Magnetic Resonance Spectroscopy (MRS), 486488,
487f, 488t
Magnetic susceptibility, 407
Magnetic tape, 114
Magnetism, 403
Magnets. See Magnetic Resonance basics, magnets
Magnets
air core magnets, 404, 404f
bipolar gradient field, 405
electromagnet, 404
FOV, 405
gradient coils, 405
gradients, coil pairs, 405, 406f
internal superconducting magnet, 405, 405f
quench, 405
RF coils, 405
shim coils, 405
superconducting air-core magnet, 405, 405f
superconductivity, 404, 405
Magnification
blurring, 209f
definition, 207
edge gradient, 209
Magnification in nuclear imaging, 683685, 690, 691f

Magnification modes, fluoroscopy, 294296
Magnification techniques, mammography,
257258
Mammography, 56, 6f
accreditation and certification, 277
anode, 241
automati, 251252
average glandular dose, radiation dosimetry, 274
backup timer, 252
breast digital tomosynthesis, 272273
Bucky factor, 256257
cathode, 240241
collimation, 250
compression of breast, 253255
computer-assisted detection and diagnosis, 270
contrast, 239, 239f
contrast degradation factor, 255
diagnostic procedure, 238
digital detectors and imaging, 263268
film
film and film processing, 260261
film viewing conditions, 262263
focal spot, 241244
half-value layer, 247248
magnification techniques, 257258
mid-breast dose, radiation dosimetry, 276
molybdenum, 241
quality assurance, 281
radiation dosimetry, 274276
reciprocity law failure, 252
regulatory requirements, 276281
scattered radiation, antiscatter grid, 255257
screenfilm cassette and film emulsion, 258
technique chart, 252253
tube port filtration, 244247
x-ray generator, phototimer system, 250253
x-ray tube design for, 188, 241, 242f
Mammography Quality Standards Act (MQSA), 238
Mammography, radiation protection, 858
Management of Persons Contaminated with
Radionuclides, 905
Manifest illness stage, acute radiation syndrome,
785786, 786f, 790, 791t, 792f
Mass attenuation coefficient, 4748, 48f,
946t954t
Mass defect, 31
Mass energy absorption coefficient, 54
Mass energy transfer coefficient, 53
Mass number, 27
Matching layer, ultrasound, 516517
Maximum intensity projection (MIP)
magnetic resonance imaging, 466467, 466f, 467f
Mean, statistics, 8182, 81f, 667
Mean life, radiopharmaceutical dosimetry, 621
Median, statistics, 80, 80f, 667668
Medical internal radiation dosimetry (MIRD) method,
of radiopharmaceutical dosimetry, 617625,
618f
MIRD formalism, internal dosimetry, 617
radiopharmaceutical doses
Memory, computer, 7072, 71f
Index
1019
Metastable state, 28
nuclear transformation, 589
Microcephaly, in utero irradiation, 826
Minification gain, output phosphor, 236
MIRD. See Medical internal radiation dosimetry
Mirror image artifact, ultrasound, 566f, 567
Mobile fluoroscopy, 303
Mobile radiography and fluoroscopy systems,
Mode, statistics, 80, 80f
Modem, 142
Modulation transfer function (MTF)
Fourier transform, 991995, 992f994f
and image quality, 270271, 270f, 271f, 272f
limiting resolution, 72, 72f
Nyquist frequency, 7374, 73f, 74f
practical measurement, 72
presampled MTF, 7475
RECT input function, 7576, 75f
resolution test phantoms, 76, 76f
SINC function, 75, 75f
Molybdenum breakthrough. See Mo-99/Tc-99m
radionuclide generator
Molybdenum-99, decay schemes, 580,
581t, 591f
Momentum, 916
Monitor pixel formats, 125126
Monte Carlo dose computation, 382383
Mo-99/Tc-99m radionuclide generator, 603604
molybdenum breakthrough, 607
MR. See Magnetic Resonance basics, Magnetic
Resonance Imaging
MR signal localization
frequency encode gradient, 441442,
442f, 443f
Gradient Sequencing, 444, 444f
magnetic environment, 438, 438f
magnetic field gradients, 438439, 438f,
439f, 439t
phase encode gradient, 443444, 443f
slice select gradient, 440441, 440f, 442f
MTF. See Modulation transfer function
Multichannel analyzers (MCAs) system, pulse height
spectroscopy, 653654, 654f
Multi-channel encoding coils, 490
Multienergy spatial registration, 685, 685f
Multi-path reflection artifact, ultrasound,
566f, 567
Multiple detector arrays, x-ray computed tomography,
313314, 317318, 329333, 314f, 317f,
318f, 329f, 330f, 331f, 332f
Multiple exposure, rating chart, 142
Multiplication factor, nuclear reactor, 600
Multiply damaged sites (MDS), DNA, 754
Multirow detector CT scanners (MDCTs), radiation
protection, 858, 865867
Multislice data acquisition, magnetic resonance
imaging, 431
Multitarget model, cell survival curves, 764765,
764f, 766f
Mutual induction, 925926, 926f
1020
Index
N
Narrow-beam geometry, 48, 48f
National Electrical Manufactures Association
(NEMA),specifications, scintillation
camera,694
Nationwide Evaluation of X-Ray Trends (NEXT), 879
Near field, ultrasound beam, 520521, 521f
Negative predictive value (NPV), 99
Negatrons. See beta minus particle and beta minus
decay
Neurovascular radiation syndrome, 790
Newtons laws, 915
Noise, and image quality, 524f, 562
Nonionizing radiation, 22
NRC. See Nuclear Regulatory Commission
Nuclear binding energy, 31
Nuclear chain reaction, 599, 600f
Nuclear fission, 31, 32f
nuclear reactor produced radionuclide, 598599
Nuclear forces, atom, 2728
Nuclear fusion, 32
Nuclear magnetic moment, 407
Nuclear magnetic moment, 407
Nuclear magnetic resonance (NMR), 402
Nuclear medicine, 880892
distance, exposure, 880882, 881t
exposure time, 880, 880t
patient protection, 887891, 889t
radioactive material spills, 887
radioactive waste disposal, 891892
radionuclide generator, 607t
shielding, 882887
Nuclear medicine planar images, 11, 12f
Nuclear reactor, 598603
Nuclear reactor-produced radionuclide
chain reaction
containment structure, 602
fast neutron, 599
meltdown, 600
multiplication factor, 600
uranium 235, 598, 599f
fission, 598599
neutron activation, 602603, 603t
Nuclear Regulatory Commission (NRC), 629, 896t
Nuclear stability, 28, 30, 30t, 29f
Nuclear terrorism, radiation protection, 903904, 904f
Nuclear transformation, 582593
alpha decay, 583584, 583f
beta-minus decay
negatron, 584585, 584f, 585f
beta-plus decay
positron emission, 585587, 586f
decay schemes, 589593, 589f592f, 593t
fluorine-18, 581t, 592f
molybdenum-99, 580, 581t, 591f
phosphorus-32, 581t, 589, 590f
radon-220, 589, 590f
technetium-99m, 580, 581t, 592f
electron capture decay, 587588, 588f
isomeric transition, 588589
metastable state, 589
metastable states, 588
pure beta emitter, 590

radioactive decay, 582
radioactivity, 579
Nucleon, 27
Nuclide classification, 28, 28t
Number system
binary form, 101102, 102t
conversion, decimal form, binary form, 102, 103t
decimal form, 101
Nyquist frequency, 7374, 73f, 74f,
MRI, 444
Doppler ultrasound, 546547, 547f
O
Occupancy factor (T), radiation shielding, 861
Off-focus radiation, anode, 185, 186f, 241
Ohms law, 921
Optical coupling, fluoroscopy, 287
Optical density, light transmission relationship, 259
Optically stimulated luminescent (OSL), 845846,
845f, 846f
Optics, 285286
Orbital binding energy, 2425
Organ conversion factors, 383t
Organ dose, 384385, 384t, 955, 958t
Organ response to radiation See Tissue reactions
Organogenesis, radiation effects on the fetus, 825
OSL. See Optically stimulated luminescent (OSL)
Output phosphor, fluoroscopy, 286287, 286f
P
PACS. See Picture archiving and communication
system
Packet, 134
Pair production, 44, 45f
Pancake probe, radiation protection, 851
Par speed system, 258
Parallel beam geometry, x-ray computed tomography,
316
Parallel-hole collimator, 678, 678f
Paralyzable model, of pulse mode, for radiation
detection, 635, 635f
Paramagnetic materials, 928
Parent nuclide, 30
Partial saturation, 421
Partial volume effect, 745
x-ray computed tomography, 368369, 369f
Particulate radiation, 2224, 23t
Path length, particle, 35
PEG. See Phase encode gradient
Period, 21
Peripheral angiography suites, 302303
Personnel dosimetry
direct ion storage dosimeter, 846, 847f
dosimeter placement, 847848
dosimeter use, 847
film badge, 843844, 844f
pocket dosimeter, 848849, 849f, 850t
problems with, 849850
protective aprons, 848
thermoluminescent dosimeter (TLD), 845846,
845f, 846f
PET. See Positron emission tomography

PET/CT imaging
artifacts
attenuation correction, 737738, 738t
PET/MRI system, 741742
Phase encode gradient (PEG), magnetic resonance
imaging, 443444, 443f
signal localization, 455, 455f
Phased array coils, magnetic resonance imaging
489490
Phased array transducer, ultrasound, 518519, 519f,
559, 559f
Phosphorus-32, decay schemes, 581t, 589, 590f
Photodiode
computed tomography, 329, 330f, 645
scintillation detector, 647
TFT array, 226, 265266, 266f
Photoelectric absorption. See Photoelectric effect
Photoelectric effect, 20
Photoelectron, 4144, 43f, 44f
Photomultiplier tube (PMT)
scintillation camera, 676, 676f
scintillation detector, 645, 647, 647f
Photon, 18
Photopeak, 656, 656f
Photopeak efficiency, scintillation camera, 686
Photostimulable phosphor, 648
Photostimulable storage phosphor (PSP),
214, 871
Phototimer system
exposure timing, 198
mammography, 250253
Physical constants, 938t939t
Picture archiving and communications systems
(PACS), 143144, 144t, 209
ambient viewing conditions, 156
challenges, 156157
DICOM, 147148
digital images, 124t, 145146, 146f
display workstations, 154156
hospital information system, 148149
IHE, 149
image display, 154
image storage, 150154, 152t, 153f
image transfer, 146147
radiology information system, 148149
teleradiology, 145
Piezoelectric crystal, ultrasound, 513515, 514f
Pincushion image distortion, in fluoroscopy,
298299
Pinhole camera, focal spot size, 183, 185f
Pinhole collimator, 678f, 679
Pocket dosimeter, 848849, 849f, 850t
Point spread function (PSF), 6163, 61f, 62f
Poisson distribution, 8485, 84f, 86t
Polyenergetic x-ray beam, 51
Portable ionization chamber, 851852
radiation protection, 642f
Positive predictive value (PPV), 99
Index
1021
Positron emission tomography (PET), 1214, 14f,

597, 597f
annihilation coincidence detection, 721723, 723f
attenuation and attenuation correction, 732734,
733f, 734f
clinical importance, 722interaction timing and
coincidence detection, 726
photomultiplier tubes, 723725, 724f, 725t, 726f
quantitative imaging, 744746
radiopharmaceuticals, 742743
random coincidence, 723, 723f
scatter coincidence, 723, 723f, 727
SPECT, compared, 743744, 744f
time of flight, 730
transverse image reconstruction, 730
true coincidence, 723, 723f
true vs. random coincidences, 726727
two-and three dimensional data acquisition,
727730, 728f, 729f
Positrons, 22, 722723
Potential energy, 916
Power Doppler, 553554, 554f
Prefixes, units, 938t939t
Preimplantation, 824
Pressure, ultrasound, 501f, 504506, 506t
Primary barrier, radiation protection, 861862
Primary x-rays, 375
Primordial radionuclides, 838
Probability density functions, 588
Procedural pause, fluoroscopy, 308
Prodromal stage, acute radiation syndrome, 785,
786f, 787t
Projection imaging, 5
Projection radiography, 207209, 207f209f
absorption efficiency, 225226, 225f
conversion efficiency, 225226, 225f
densitometer, 261
latitude, 213214
overall efficiency, 244
Promotion, cancer development stages, 794
PROPELLER, magnetic resonance imaging,
457, 458f
Proportional counter, gas-filled radiation detector, 641
Proportional region, gas-filled radiation detector, 639
Protective aprons, 848
Protocol, computer network, 134
Proton, magnetic resonance imaging, 409410, 410f
Pulse echo operation, ultrasound, 10, 10f
Pulse height analyzers (PHAs), 651
multichannel analyzer, 653654, 654f
single channel analyzer, 652653, 652f, 653f
Pulse height spectroscopy, 651659
cesium-137, spectrum of, 655657, 656f
iodine-125, spectrum of, 657658, 658f
technetium-99m, spectrum of, 657, 657f
Pulse mode, radiation detection, 634635, 635f
non-paralyzable mode, 635, 635f
paralyzable model, 635, 635f
Pulse repetition frequency (PRF), ultrasound, 530, 530f
Pulse repetition period (PRP), ultrasound, 530, 530f
1022
Index
Pulse sequences, Magnetic resonance imaging

contrast-weighted images, 421
gradient echo
coherent GE, 432434, 433f, 434t
FID signals, 431
flip angle, 431, 342f
GE acquisition, 436
GE sequences, 437, 437f
incoherent GE, 434436, 435f
Mxy signal, 431
rotating frame, 428f, 431f
short TR, 432, 433f
spoiled GE, 434436, 435f
steady-state free precession (SSFP), 436, 436f,
437f
inversion recovery
FLAIR, 429430, 430f
RF sequence, 428, 428f
steady-state equilibrium, 427
STIR, 428429, 428f, 429f
spin echo
contrast weighting, 422423
90-degree pulse, 422, 422f
180-degree RF pulse, 422, 423f
proton density weighting, 424425, 425f, 427t
SE parameters, 424, 427f, 427t
T1 weighting, 423, 424f
T2 weighting, 425426, 426f
Pulsed fluoroscopy, 293294, 294f
Q
Quality control
dose calibrator, 665666, 666f
mammography, 260262
PET, 735
scintillation, 695697, 696f, 697f
SPECT, 13, 717721
thyroid probe, 663
well counter, 663
Quality factor, ultrasound, 516
Quanta, 18
Quantization, data conversion, 108
Quantum detection efficiency (QDE), 225, 291f
radiation detection, 637
Quantum levels, electron, 24
Quantum limited detector., 219
Quantum mechanics model, 413
Quantum noise, 80
Quantum number, electron shell, 24
R
Rad
definition, 54
roentgen to rad conversion factor, 274
RADAR formalism, internal dosimetry,
618620, 619f
Radiation
absorption, 18
electromagnetic radiation, 1822
electromagnetic spectrum, 19f
electromagnetic wave characteristics, 21, 21f
electron interaction
bremsstrahlung radiation, 37
elastic, 36
inelastic, 37
electron transitions, 2627
frequency, 21
Gamma ray, 20, 30
interaction with matter, 18
attenuation, electromagnetic radiation,
3844
linear attenuation coefficient, 4547, 46f
mass attenuation coefficient, 4748, 48f
particle interactions, 3338
internal conversion electrons, 30
ionization, defined, 633
ionizing radiation, 22
leakage, 187
mass-energy equivalence, 2324
neutron interactions, 38, 38f
particle characteristics, 2122
photon, 21
quanta, 18
particle interaction
annihilation, 726
Bragg peak, 35
delta rays, 33
excitation, 3337, 34f
ionization, 3437, 34f, 35f
linear energy transfer, 36
radiative loss, 3337, 37f
secondary ionization, 33
specific ionization, 3435, 35f
particulate radiation, 2224, 23t
photon, 18, 22
quanta, 18
quantities and units
fluence, 52
flux, 52
gray, 53
kerma, 5253
rad, 54
rem, 57
roentgen, 55
sievert, 56
radio frequency, 18
scattering, 18
types of, 18
uncharged particle interaction, 36
wavelength, 21
wave-particle duality, 18, 20, 20f
weighting factor, 56
x-ray, 18, 26
Radiation biology
acute radiation syndrome, 785791
biological effects
classification of, 752
determinants of, 751752
deterministic, 752
stochastic, 752
cellular radiobiology
adaptive response, 770771
bystander effect, 771772

cell survival curves, 764765, 764f, 766f
cellular radiosensitivity, 766770, 770t
genomic instability, 772
deterministic effects. See Tissue reactions
hereditary effects of radiation exposure, 821823
organ response to high dose. See Tissue reactions
radiosensitivity, 774, 774f, 775t
regeneration/repair, 773, 773f
reproductive organs, 777, 781782
skin, 774777, 778t, 779f, 780f, 781t
threshold doses, adults, 783t, 784t
radiation-induced carcinogenesis, 788818
age and gender risk estimates, 811814,
811t812t, 813f, 814f
age as risk factor, 797798
BEIR VII report, 806
breast cancer, 819821, 820t
cancer developmental stages, 794
dose rate and fractionation, 797
dose-response models, 804809, 807f, 808f
environmental risk factors, 794795
epidemiologic investigations, 799802,
800f801f, 801t, 803t, 804t
genetic susceptibility, 798799
incidence rate, 791792
leukemia, 816, 818f
lifetime risk estimate, cancer incidence and mortality, 792, 793t
low-dose, low-LET radiation, 802, 805806
molecular biology and cancer, 793794
multiplicative and additive risk models,
809811, 810f, 810t
population radiation risk estimates, 814816,
815f, 815t, 817f
radiation quality, 796797, 797f
risk expressions, 795796
thyroid cancer, 817819
radiation-induced DNA damage and response
chromosomal aberrations, 762763, 763f
DNA damage spectrum, 757760, 759f, 760f
DNA repair, 761760, 761f
tissue/radiation interaction, 752757, 753f,
756f, 757f
in utero, radiation effects
gestation, 824831
radionuclides, 828, 830831
Radiation detection
anions, 633
cation, 633
counters, 634
current mode, 634635, 635f
dead-time, 634
detector types, 633637
dosimeter, 634
efficiency, detection, 636637, 637f
energy-compensated detectors, 642
exposure rate, 639
gas-filled radiation detector, 637643
Geiger-Mueller counter, 641643, 642f
ionization chamber, 639641, 640f
proportional counter, 641
Geiger-Mueller region (GM region), 639
Index
1023
ionization chamber region, 639

proportional region, 639
pulse height spectrum, 635636, 636f
pulse mode, 634635
non-paralyzable model, 635, 635f
paralyzable model, 635, 635f
recombination region, 639
reverse bias, 647
scintillation detector. See Scintillation detector
semiconductor detector, defined, 651
spectrometers, defined, 634
spectroscopy, defined, 635, 636f
systematic error, 667
thermoluminescent dosimeter, 648
TLD, 648
Radiation dose
annual cosmic radiation doses, 998, 1003f
vs. effective dose, 998, 1003f
emission of radiation, 998, 1000f
estimated effective doses vs.standard myocardial
ICRP tissue weighting factors, 998, 999t
imparted energy, 5556
life expectancy, 998, 1002t
natural and artificially produced sources,
998, 1003t
organ equivalent doses, 998, 1000f
public and occupational radiation exposure, 998,
1002t
Radiation Dose Structured Report (RDSR), 875
Radiation protection
accident, 902903
acute radiation syndrome, 906907
advisory bodies
International Commission on Radiological Protection, 893
National Council on Radiation Protection and
Measurement, 893
bone mineral densitometry x-ray system, 859
by-product material, 892
capita effective dose, 840f
collective effective dose, 840, 840t
committed dose equivalent, 894
computed tomography, patient protection,
874877, 875t, 876t
consumer products and activity, 841
contamination control and survey, 883, 885
controlled and uncontrolled areas, 855
decay chain, 838
dental and veterinary x-ray facilities, 859
design goals, 855, 856
diagnostic and interventional X-ray imaging
patient protection, 869879
personnel protection, 867869, 868f, 870f
dose limit, 894
dose optimization, 876877
effective dose equivalent, 894
equipment, 850852
Geiger-Mueller survey instruments, 850851
portable ionization chamber, 851852
error prevention, 897899
exposure control. See Exposure control
1024
Index
Radiation protection (Continued)

exposure rate constant, 881, 883t
external radioactive contamination, 904905
fluoroscopic imaging systems, 857
fluoroscopy, 872874
gonadal shielding, 871
imaging of pregnant patients, 901902
intake and derived air concentration, 894895
internal radioactive contamination, 905, 906t
leaded glasses, 869
leakage radiation, 856, 856f
mammography, 858
management, 899900
management of persons contaminated with
radionuclides, 905
MDCTs, 858
medical events, 887
mobile radiation barrier, 869
mobile radiography and fluoroscopy systems,
858859
natural radiation, 839, 841
nuclear medicine, 880892
distance, exposure, 880882, 881t
exposure time, 880, 880t
patient protection, 887891, 889t
radioactive material spills, 887
shielding, 882887
occupational exposure, 841842
optically stimulated luminescent dosimeters,
845846, 845f, 846f
pancake probe, 851
patient protection, 869
tube voltage and beam filtration, 870
x-ray image receptor, 871872
personnel dosimetry
direct ion storage dosimeter, 846, 847f
dosimeter placement, 847848
dosimeter use, 847
film badge, 843844, 844f
pocket dosimeter, 848849, 849f, 850t
problems with, 849850
protective aprons, 848
thermoluminescent dosimeter (TLD), 845846,
845f, 846f
PET, 842
PET/CT and SPECT/CT shielding, 867
photostimulable storage phosphor (PSP), 871
primordial radionuclides, 838
protective gloves, 869
radiographic room, 860f, 862, 863f
radiological and nuclear terrorism, 903904,
904f
radionuclide therapy, 888891
radon-222, 838, 839
RDSR, 875
regulatory agencies, 892893
safety culture, 898
scattered radiation, 856, 856f
secondary barrier, 862
sentinel event, 898
shielding barrier calculation

CT scanner, 865867, 865f, 866f
radiographic room, 862864, 863f,
survey, 864865
shielding
terminology, 860862, 860f, 861t
workload distribution, 860861
special nuclear material, 892
Radiation Risks of Diagnostic Imaging, 898
Radiation weighing factor, 56
Radiation-induced arteriole fibrosis, 774f
Radio nuclide
production, 594608
Radioactive decay, 30, 579, 581t, 582
activity, 579
decay constant, 579580
fundamental decay equation, 582
physical half-life, 580581, 581t
Radioactivity, 28, 30
Radiofrequency (RF) coils, 405
Radiography, 45, 4f, 5f
absorption efficiency, 225226
CCD chip
amount of light, 218
dexel, 217
slot-scan X-ray systems, 219
X-ray quantum limited detector, 219
CMOS, 219220
computed
BaFBr, 214
F-center, 215, 216
imaging plates, 214
laser light stimulation, 214
PSP, 214
readout mechanics, 215f
storage phosphor, 214
conversion efficiency, 225226
detectors
histogram values, 227
speed, 213
dose-related metrics, 377382
dual-energy
attenuation characteristics, 229
chest radiography, 229
Compton scattering component, 229
logarithmic weighted subtraction, 230
geometry of projection
beam divergence, 207
detector, 207
SID, 208
intensifying screens, 210, 224225
proprietary exposure index (EI), 228
scatter reduction, 235236
scattered radiation
anti-scatter grid. See Anti-scatter Grid
scatter-to-primary ratio, 231
X-ray, 230
scintillators, 224225
screen-film. See Screen-film radiography
skin dose, 380
technique factors, 223224
Radiology information system (RIS), 148149

Radionuclide generator
Mo-99/Tc-99m radionuclide generator, 603604
nuclear medicine, 607t
quality control, 607608, 666
secular equilibrium, 606, 608f
Sr-82/Rb-82 generator, 666667
transient equilibrium, 604, 606, 606t
Radionuclide production, 594608
containment structure, 601
control rods, 600
cumulated activity, 620
cyclotron-produced radionuclide, 594598, 595f,
597f
effective half-life, 620
fission fragments, 598
fuel rods, 600
generator. See Radionuclide generator
mean life, 621
meltdown, 600
molybdenum breakthrough, 607
multiplication factor, 600
neutron activation, 601
nuclear reactor-produced radionuclide, nuclear
fission 606611,607f
reactor, 600, 601f
S factor, 618
supercritical, 600
thermalize, 599
time-integrated activity, 618
U.S. Pharmacopeia (USP), 608
Radionuclide therapy, 10051007
Radiopharmaceuticals, 608616
absorbed dose, 976t977t
administration methods, 956t967t
adult dose, 968t971t
authorized user, 629
breast dose, 981t
breast-feeding, 984985
byproduct material, 629
clinical utility, 956t967t
dose from, 743
effective dose, 972t973t
ideal diagnostic
high target/nontarget activity, 609, 613
low radiation dose, 609
safety, 613
internal dosimetry, 617
localization, 956t967t
localization mechanisms, 613616
medical events, 630631
medical imaging, 608609
NDA, 628
physical characteristics of, 610t612t
safety, 613
U.S. Food and Drug Administration, regulation, 628
written directive, 629631
Radon-220
decay schemes, 589, 590f
Radon-222, 838, 839
Index
1025
Random error, counting statistics, 667

Random number generator, 382
Range, particle, 35
Rayleigh scattering, 3839, 39f
RDSR. See Radiation Dose Structured Report (RDSR)
Readout gradient, 441
Receive focus, ultrasound beam, 522, 522f
Receiver operating characteristic (ROC) curves, and
image quality, 9699, 96f, 97f, 99f
Reciprocity law failure, mammography, 252
Recovery stage, acute radiation syndrome, 785, 786f
Rectifier circuit, x-ray, 192
Reference axis, x-ray beam, mammography, 243
Reference levels, 878879
Reflection, ultrasound, 508509, 508f, 509t
Refraction, ultrasound, 508f, 509510
Refraction artifacts, ultrasound, 563, 564f
Relative biological effectiveness (RBE), 756
Relative risk (RR), radiation biology, 795
Remote fluoroscopy rooms, 301302
Resonance transducers, ultrasound, 515, 515f
Reverberation, ultrasound, 563564, 565f
Reverse bias, photodiodes, 647, 650
RF coil safety, magnetic resonance imaging, 491
RF receiver coil, 489
RF transmitter coils, 489
Road mapping, fluoroscopy, 297
ROC. See Receiver operating characteristic
Roentgen
definition, 4
in exposure rate constants, 882883
to rad conversion factor, 274
Roentgenography, 45, 4f, 5f
Rotating anode, 180
Rotating frame, magnetic resonance imaging, 411
Rotor/stator, x-ray tubes, 176, 180182, 180f, 181f
Routers, 140
S
S distortion, fluoroscopy, 290
Sampling
analog-to-digital converters, 108
image quality and, 62, 64, 64f
Sampling pitch
computed tomography, 338
mammography, 265
Scatter (contrast) degradation factor
mammography, 255
Radiography, 235
Scattered radiation
antiscatter grid
mammography, 255257
in nuclear imaging, 681
in projection radiography system, 207209,
207f209f
radiation protection, 856, 856f
scatter coincidences, 727
Scattered x-rays, 375
Scintillation camera
collimator, 677679, 678f
efficiency, 688690, 689f, 690t
resolution, 688690, 689f, 690t
1026
Index
Scintillation camera (Continued)

design, 686694
collimator resolution, and collimator efficiency,
688690, 689f, 690t
detector, 676677, 676f
electronics, 676677, 676f
intrinsic spatial resolution, and intrinsic efficiency, 686688, 687t
spatial linearity, and uniformity, 691694, 692f,
693f
system spatial resolution, and efficiency,
690691, 691f
energy resolution, 686
image formation principles, 680681, 680f
intrinsic efficiency, 686687, 686688, 687t
intrinsic resolution, 683
intrinsic spatial resolution, and intrinsic efficiency,
686688, 687t
multienergy spatial registration, 685, 685f
multiple window spatial registration, 685
National Electrical Manufactures Association
(NEMA), specifications, 694
operation/quality control, 695697,
696f, 697f
performance, 682686
measures, 682686, 683f685f
photopeak efficiency, 686
radiation, 18, 36
region of interest (ROI), 701
scintillation camera, 674704
spatial linearity, 691694, 692f, 693f
spatial resolution, 682
system efficiency, 685
uniformity, 691694, 692f, 693f
Scatter-to-primary ratio (SPR), 231
Scintillation detector, 643648
afterglow, defined, 643
excited electron, 647648
photostimulable phosphor, 648
thermoluminescent dosimeters, 647
inorganic crystalline scintillator, 644645, 646t
luminescence, 643
photodiode, 647
photomultiplier tube, 645, 647, 647f
photostimulable phosphors, 648
Scintillators, 224225
Screen-film combination, screen-film
system, 258
Screen-film radiography
characteristic curve
H and D, 212
film contrast, 213
intensity of light, 213
latitude, 214
double-sided screen-film, 210f
film processing, 211212, 212213
grain shape, 211
indirect detection process, 211f
latent image center, 211
Screen-film systems
projection radiography, 209
Screening mammography, 238
Screens
composition and construction, 210
intensifying, 224225
Secondary barrier, radiation protection, 862
Secondary ionization, 33
Secondary quantum sink, 219
Sector scanning, ultrasound imaging, 10
Secular equilibrium, 606, 608f
SEG. See Slice encode gradient
Segmentation, three-dimensional image display, x-ray
Selectivity, grids, 235
Self-induction, 924925
Self-magnetism, 407
Semiconductor detector, radiation detection,
648651, 649f, 650f
Sensitivity speck, film processing, 211
Sensitometer, 261, 261f
Sentinel event, radiation protection, 898
Septa, scintillation camera collimator, 677
Server, 133
Shadowing artifact, ultrasound, 563, 564f
Shielding
barrier calculation, 866867, 866f
calculation, 862864, 863f
CT scanner, 864866, 865f, 866f
diagnostic radiology, 854
exposure control, 854
installation, 864
nuclear medicine, 882887
PET/CT and SPECT/CT, 867
Shift invariant, point spread function, 62
Shim coils, magnetic resonance imaging, 405
SID. See Source-to-Imager Distance
Side lobe artifact, ultrasound, 523f, 524f, 565f, 566567
Side lobe, ultrasound beam, 523, 523f, 524f
Sievert, 56
Signal averages, magnetic resonance imaging, 461
Signal-to-noise ratio (SNR), 9192, 92f
Similar triangles, basic geometric principles, 207
Simple backprojection, x-ray computed tomography,
351352
Simple kinetic model, internal dosimetry, 620, 621f
Single channel analyzers (SCAs), pulse height
spectroscopy, 652653, 652f, 653f
Single exposure, rating chart, 200
Single film emulsion, single screen,
mammography, 258
Single phase x-ray generator, 195, 197, 200
Single photon emission computed tomography
(SPECT), 12, 13f
attenuation correction in, 712713, 712f
axis of rotation (AOR), 717718, 718f
conventional planar imaging, compared, 716
coronal and sagittal image generation, 713
head tilt camera, 719720, 719f
image acquisition, 706707, 709, 709f
multienergy spatial registration, 717
multihead cameras, 714715
oblique image generation, 713714
phantoms, 720, 720f, 721t

positron emission tomography, compared,
743744, 744t
principle, 706, 707f, 708f
quantitative imaging, 744746
spatial resolution, 715, 716f
transverse image reconstruction, 709711, 710f,
711f, 712f
uniformity, 718719, 718f
X and Y magnification factors, 717
Sinogram, tomographic reconstruction, 355f, 721
Skin dose, 380382
Skin, response to radiation, 774777, 778t, 779f,
780f, 781t
Slice encode gradient (SEG), magnetic resonance
imaging, signal localization, 459
Slice sensitivity profile, x-ray computed tomography,
360, 361f, 362f
Slice thickness
contrast resolution, x-ray computed tomography, 366
single detector array scanners, computed
tomography, 333334
spatial resolution, x-ray computed tomography, 366
ultrasound, 567568, 568f
Slip-ring technology, helical scanner, x-ray computed
tomography, 372
Slit camera, focal spot x-ray tube, 183
Smoothing, spatial filtering, computer image
processing. See Spatial filtering
SNR. See Signal-to-noise ratio
SOD. See Source-to-object distance (SOD)
Sodium iodide scintillator, 676
Sodium iodide thyroid probe/well counter, 660663,
660f, 662f
Sodium iodide well counter, 661662, 662f
Software
applications programs, 112, 141, 937
computer languages
high-level languages, 937
machine language, 937
operating system, 937
Solid-state detector, x-ray computed tomography, 329
Sound. See Ultrasound
Source to imager distance (SID), 208
Source-to-object distance (SOD), 871
Source-to-skin distance (SSD), 377378, 379f
Space charge effect
cathode, 178f
mammography, 250
Space charge limited, cathode, x-ray tubes, 178f
Spatial filtering, nuclear imaging, 701
Spatial resolution, 298299, 299f
convolution, 994
Fourier transform, 991, 993f, 995
image quality, 6065
line spread function, 730
ultrasound beam, 524527
x-ray computed tomography
Spatial resolution,
line spread function, 689f
Index
1027
Specific ionization, 3435, 35f

SPECT. See Single photon emission computed
tomography
Spectroscopy
defined, 635636, 636f
radiation detection, 635636, 636f
SPECT/X-ray CT system
myocardial perfusion imaging, 741
Speed artifact, ultrasound, 565f, 566
Speed, of sound, 501f, 502504, 503t, 505f
Spin echo, magnetic resonance imaging, 422
Spin-lattice relaxation, magnetic resonance imaging,
418
Spin-spin interaction, magnetic resonance imaging,
417f
Spiral filling method, magnetic resonance imaging,
456f, 457
Spoiled gradient echo technique, magnetic resonance
imaging, 434436
Spurs, 753
Standard deviation, image quality, 363364
Standardized uptake value (SUV), PET imaging, 745
Star pattern, focal spot size, 183, 184, 186f
Static, frame mode acquisition, image acquisition,
699700, 699f
Stationary x-ray tube anode, 180
Stationary invariant, point spread function, 62
Stereotactic breast biopsy, 270271
Stochastic effect, radiation-induced cancer, 752
Storage of images
data compression, 152154
image management, 151152
schemes, 150151
Storage phosphors, 214
Structure, atom. See Atom structure
Structure noise, 78f, 79
Subject contrast, and image quality, 8788, 88f
Sum peak, radiation spectrum, nuclear imaging, 658
Superconductivity magnets, magnetic resonance
imaging, 404, 405
Surface coils, magnetic resonance imaging, 489
Surface rendering, three-dimension image display,
Susceptibility artifacts, magnetic resonance
imaging,407
System efficiency, scintillation camera, 685
Systematic error, counting statistics, 667
Systeme International (SI), 914
T
T1 weighting, magnetic resonance imaging, 423, 424f
T2 weighting, magnetic resonance imaging,
425426, 426f
Target exposure index, 228
TCP/IP. See Internet Protocol Suite
TDI. See Time delay and integration
Technetium-99m
decay schemes, 580, 581t, 592f
spectrum of, 657, 657f
Technique chart, mammography, 252253
Teleradiology quality control, 157159
Temporal resolution, 300301
1028
Index
Teratogenic effects, in utero irradiation, 825827

TFT. See Thin-film transistor
The Joint Commission, 897899
Thermionic emission cathode, 177, 178f
Thermoluminescent dosimeter (TLD), 648
radiation detection, 648
radiation protection, 845846, 845f, 846f
Thin-film-transistor (TFT)
displays, 220
flat panel radiographic detector, 220222
Three-dimensional Fourier transform imaging, signal
localization, 459460, 459f
Three-phase x-ray generator, 195
Threshold, radiation exposure, 807, 807f
Thyroid cancer, radiation induced carcinogenesis,
817819
Thyroid probe-well counter, sodium iodide, 660663,
660f, 662f
Thyroid shields, radiation protection, 869
Thyroid uptake measurements, 661
Time delay and integration, 219
Time of echo (TE), magnetic resonance imaging,
420421
Time of inversion (TI), magnetic resonance imaging, 421
Time of repetition (TR), magnetic resonance
imaging,420
Time-activity curve (TAC), scintillation camera, 701
Tissue weighting factors, 384t
Tissue reactions, radiation 772784, 783784t
ocular effects, 782783
radiosensitivity, 774, 774f, 775t
regeneration/repair, 773, 773f
reproductive organs, 777, 781782
skin, 774777, 778t, 779f, 780f, 781t
threshold doses, adults, 783784t
Tissues magnetization properties
free induction decay
amorphous structure, 416
antenna coil, 415
inhomogeneities, 416
magnetic field domains, 416
phase coherence, 416
sinusoidal electronic signal, 415, 416f
T2 relaxation time, 416, 417f
longitudinal magnetization equilibrium, 417
spin-lattice relaxation, 417, 417f, 418f, 419f
T1 relaxation time, 418
tumbling frequencies, 418, 419f
T1 and T2 comparison, 419420, 419f, 420t
TLD. See Thermoluminescent dosimeter (TLD)
Tomographic emission images, 12, 13
Tomographic reconstruction. See Image reconstruction
Transducer arrays, ultrasound, 513519, 513f, 514f,
515f, 516f, 527
Transducer, ultrasound, 513519, 513f, 514f, 515f, 516f
Transformers, x-ray, 191194, 192f
Transient equilibrium, 604, 606, 606t
Transmission Control Protocol/Internet Protocol. See
Internet Protocol Suite
Transmission imaging, 5
Transmit focus, ultrasound beam, 2223, 23f
Transmit/receive switch, ultrasound system, 529
Transmittance, optical density, film, 259

Transverse magnetization, 412
True-positive fraction (TPF), 97
Tube output, mammography, 248250
Tube port and filtration mammography, 244247
Tube, x-ray. See X-ray, tube
Tumor suppressor genes, 793
Tungsten atom, de-excitation of, 27f
Two dimensional data acquisition, magnetic resonance
imaging, 450451, 450f
Two-pulse waveform, x-ray generator circuit
designs, 197
U
Ultrasound, 500576. See also Doppler ultrasound
absorption, 507
acoustic impedance, 507, 507t
acoustic power, pulsed ultrasound intensity,
572574, 573f, 573t
ambiguity artifact, 567
A-mode transducer, 533f, 534535
analog to digital conversion, 531, 532f
array transducer, 521524, 522f
artifacts, 560568
attenuation, 511513, 511t, 512f
axial resolution, and image quality, 524525, 525f
beam former, 529
beam steering, 532, 532f
biological mechanism, effect, 574575, 576f
B-mode, transducer, 536, 537f
characteristics 501506
contrast sensitivity, noise, and image quality, 562
damping block, 515516, 516f
decibel, 506, 506t
demodulation and envelope detection, 534
Doppler performance measurements, 571
dynamic focusing, 532, 532f
dynamic range, 534
electronic scanning and real-time display, 537539,
538f, 539f
elevational resolution, and image quality, 524f,
527, 527f
enhancement artifact, 563, 564f
frequency, 501f, 502504, 503t, 505f
grating lobe artifact, 523f, 524f, 565f, 566567
harmonic imaging, 556559, 557f, 558f
image data acquisition, 527536
image quality, 560568
intensity, 501f, 504506, 506t
interaction with matter, 506513
lateral resolution, and image quality, 525526, 526f
logarithmic amplification, 534
longitudinal wave, 502
matching layer, 516517
measurements area, distance, volume, 554555, 555f
mechanical energy, propagation, 501502, 501f
mechanical scanning and real-time display, 537
mirror image artifact, 566f, 567
M-mode, transducer, 535, 535f
multipath reflection artifact, 566f, 567
phased array, 519
piezoelectric crystal, 513515, 514f

preamplification, 531, 532f
pressure, 501f, 504506, 506t
pulse echo operation, 529531, 530f, 531t
pulse echo technique, 507, 507t
pulser, 529
quality assurance, 567f, 568571, 568f, 571t
receiver, 532534, 533f
rectification, 534
reflection, 508509, 508f, 509t
refraction, 508f, 509510
resonance transducers, 515, 515f
reverberation, 563564, 565f
scan converter, 536
scattering, 510511, 510f
shadowing artifact, 563, 564f
side lobe artifact, 523f, 524f, 565f, 566567
signal summation, 532, 532f
slice thickness, 567568, 568f
sound, characteristics of. See Sound, characteristics of
spatial compounding, 539
spatial resolution, 527f, 535f, 561562
special purpose transducer assemblies, 559
specular reflector, 510
speed artifact, 565f, 566
speed of, 501f, 502504, 503t, 505f
system performance, 567f, 568571, 568f, 571t
three-dimensional imaging, 560, 561f
transducer, 513519, 513f, 514f, 515f, 516f
transmit/receive switch, 529
two-dimensional image display and storage, 536542
ultrasound contrast agents, 555556
wave interference patterns, 505f
Ultrasound beam
far field, 521
Fraunhofer region, 521
Fresnel region, 520
grating lobes, 523, 523f, 524f
Huygens principle, 520
nearfield, 520521, 521f
properties of, 514
side lobes, 523, 523f, 524f
spatial resolution, 524527
Ultrasound imaging, 911, 10f
Uncharged particle interaction, 36
Units, radiation. See Radiation quantities
U.S. Food and Drug Administration
Radiopharmaceuticals, regulation, 628
U.S. Nuclear Regulatory Commission, 893896
radiopharmaceuticals regulation, 663
regulation, ALARA principle, 896
regulatory dose limits and comparison to ICRP 103
recommended dose limits, 896t
U.S. Pharmacopeia (USP), 608
Use factor, radiation shielding, 861862
Useful beam, sources of exposure, radiation
protection, 856
V
Variable frame rate, pulsed fluoroscopy, 293294
Varying magnetic field effects, biological effects,
Vector versus Scalar Quantities, 913914, 914f
Index
1029
Veiling glare, 122, 123, 287

Video cameras, fluoroscopy, 288289, 288f
Video interface, 117
Voltage, Bremsstrahlung spectrum, 171
Voltage ripple, x-ray, 197198, 197f
Volume coils, 489
Voxel volume
computed tomography, 325327, 325f
W
WAN. See Wide area network
Wavelength
electromagnetic wave, 21
ultrasound, 501f, 502504, 503t, 505f
Wave-particle duality, 1820
White noise, 87
Wide area network, 133, 138f
Windowing, 127128
Word, digital data representation, 105
Word mode, digital image formats in nuclear
medicine, 698
Workload, radiographic, fluoroscopic, CT
radiographic distribution, 861
Wrap-around artifacts, magnetic resonance imaging,
485486, 485f
Written directive, radiopharmaceuticals,
629631
X
Xenon detector, x-ray computed tomography, 640
X-ray
emission, factors affecting, 202206
exposure timing
electronic switch/timer, 198
falling-load generator control circuit, 198199
mechanical contactor switch, 198
phototimer switch, 198199, 199f
filtration, 189190
generator
components
autotransformers, 191194
diodes, 193
filament circuit, 177
rectifier circuit, 192
transformers, 191194, 192f
exposure timing, 176177
line voltage compensation, 196
phototimer switch, 198199, 199f
types
constant potential generator, 197, 197f, 198
inverter generator, 195, 195f
single phase x-ray generator, 195, 197, 200
three-phase x-ray generator, 195
voltage ripple, 197198, 197f
positive beam limitation, 187
production
bremsstrahlung process, 171174, 172f, 173f
x-ray spectrum, 174175, 174t, 175f,
175t, 176f
1030
Index
X-ray (Continued)
tube
anode
angle, 181f, 182184, 182f, 184f, 185f
configuration, 180182, 180f, 181f
focal spot size, 182184
heel effect, 184185, 186f
off-focus radiation, 185, 186f
pinhole camera, 183, 185f
rotating, 180182, 181f
slit camera, 183
star pattern, 184
stationary, 180182
biased focusing cup, 178179, 179f
cathode, 177179, 178f, 179f
cathode block
current, 176
getter circuit, 187
grid biased tube, 188
housing, 171, 176
mammography tube, 188
insert, 171, 177f
leakage radiation, 187
port, 176
space charge effect, 178f
thermionic emission, 177, 178f

voltage, 176
X-ray beam filters, 171
X-ray cable sockets, 176
X-ray computed tomography
detector type, 329335
fan beam geometry, 316
filtered backprojection, 352354, 353f, 354f, 355f
Hounsfield units, 324
iterative reconstruction, 357358, 357f, 358f
partial volume, 368369, 369f
simple backprojection, 351352, 352f, 353f
third-generation, 371372, 372f
X-ray dosimetry
absorbed dose, 386387
CT. See CT dosimetry
dose estimation, 400
effective dose. See Effective dose
equivalent dose. See equivalent dose
Monte Carlo Dose computation, 382383
organ doses. See Organ dose
X-ray generator, 171
X-ray spectrum, 174175, 174t, 175f, 175t, 176f
X-ray tube output, dose-related metrics, 377,
378f, 378t

Essential Physics of Medical Imaging

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Essential Physics of Medical Imaging

Transféré par

Droits d'auteur :

Formats disponibles

The Essential

JERROLD T. BUSHBERG, PhD

J. ANTHONY SEIBERT, PhD

EDWIN M. LEIDHOLDT JR, PhD

JOHN M. BOONE, PhD

Executive Editor: Charles W. Mitchell

Preface to the Third Edition

Preface to the Third Edition

Jiang Hsieh, PhD

Fred A. Mettler Jr, MD, MPH

Dianna Cody, PhD

Michael Cronan, RDMS

John Sabol, PhD

Robert Dixon, PhD

Mahadevappa Mahesh, PhD

Jeffrey Siewerdsen, PhD

Cynthia McCollough, PhD

Michael G. Stabin, PhD

Ken Eldridge, RT(R)(N)

Steve Wilkendorf, RDMS

Michael McNitt-Gray, PhD

Kai Yang, PhD

Section I: Basic Concepts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Radiation and the Atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Interaction of Radiation with Matter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Medical Imaging Informatics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

Section II: Diagnostic Radiology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

10 Computed Tomography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312

11 X-ray Dosimetry in Projection Imaging and Computed Tomography. . . . 375

12 Magnetic Resonance Basics: Magnetic Fields, Nuclear Magnetic

13 Magnetic Resonance Imaging: Advanced Image Acquisition

Section III: Nuclear Medicine. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577

16 Radionuclide Production, Radiopharmaceuticals,

17 Radiation Detection andMeasurement. . . . . . . . . . . . . . . . . . . . . . . . . . . 633

18 Nuclear ImagingThe Scintillation Camera. . . . . . . . . . . . . . . . . . . . . . . . 674

19 Nuclear ImagingEmission Tomography. . . . . . . . . . . . . . . . . . . . . . . . . . 705

Section IV: Radiation Biology and Protection . . . . . . . . . . . . . . . . . . . . . . . 749

21 Radiation Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 837

Section V: Appendices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 911

Introduction to Medical Imaging

1.1 The Modalities

Section I Basic Concepts

Chapter 1 Introduction to Medical Imaging

Section I Basic Concepts

Chapter 1 Introduction to Medical Imaging

Section I Basic Concepts

Magnetic Resonance Imaging

Chapter 1 Introduction to Medical Imaging

using specialized MRI sequences to evaluate the biochemical composition of tissues

Section I Basic Concepts

Chapter 1 Introduction to Medical Imaging

Nuclear Medicine Imaging

Nuclear Medicine Planar Imaging

Section I Basic Concepts

FIGURE 1-7 Anterior and posterior whole-body bone

Single Photon Emission Computed Tomography

Positron Emission Tomography

Chapter 1 Introduction to Medical Imaging

a nnihilation. When a photon pair is detected by two detectors on the scanner, it

Section I Basic Concepts

PET/CT combined imaging has applications in oncology, cardiology, neurology,

Combined Imaging Modalities

Chapter 1 Introduction to Medical Imaging

1.2 Image Properties

TABLE 1-1THE LIMITING SPATIAL RESOLUTIONS OF VARIOUS MEDICAL

E x-ray E b vacant shell E b transition shell

TABLE 2-2NUCLEAR FAMILIES: ISOTOPES, ISOBARS, ISOTONES,